Sony TRV900 FAQ

by John Beale (beale@best.com) v8.7 July 29 2004

[09/03/01] Here are some questions and answers regarding the Sony DCR-TRV900 three-CCD-chip, MiniDV-format digital video camera. I bought one of these cameras soon after it came out. I don't own stock in Sony. The usual disclaimers apply... my opinions are my own. Note: I'm assuming you've read the information on Sony's web page about this camera. This FAQ is intended to complement the official stuff.  Feel free to send me questions and/or answers you'd like included in this FAQ. The date in brackets before some paragraphs indicates when they were most recently updated.

I'd also like to thank the many readers of my web page and the TRV900 mailing list for sharing their ideas and pictures with me and the greater Internet community. Thanks folks!

John Beale
Fremont, CA
Sept. 2001


The following questions are answered in this FAQ:


Most frequently asked

You can download the entire TRV900 User's Manual or TRV950 User's Manual online, and there are also other manuals at www.docs.sony.com. The TRV890/TRV900 Service Manual was at one time, and may still be available here. There are some DVCAM manuals online from Sony Canada. This TRV900 FAQ addresses questions which are, for the most part, not covered in the user's manual. I bought my TRV900 mail-order from Supreme Video+Electronics (now Etronics). I specified next-day delivery, and they delivered (Sept. 9 1998). I also got the NP-F750 battery and a UV filter; see my accessories page for more. Sony's warranty is 1 year parts, 90 days labor. (In the USA, you can buy an extended warranty from Sony: call (800) 378-4590.) The online vendors I use and recommend are B+H Photo, Profeel, and Etronics in New York, CameraWorld in Oregon, One Call in Washington, and Video Direct in Florida. I have found all these places have a good reputation on the net, and I have purchased various equipment from each one of these dealers and been satisfied.

As of July 2002, the TRV900 model has been discontinued by Sony in favor of the new DCR-TRV950. I am keeping this FAQ online as a reference for current TRV900 owners, and those buying used. As of 2003, I believe it is unlikely that you would find a new, unused TRV900 anywhere but you can still find a used TRV900 on Ebay and other online auction sites. Do be aware of the risks of buying sight-unseen; check out your potential vendor as well as possible. If you buy a used camera, read Doug Graham's article on buying used (many other good articles at videouniversity.com, by the way.) See the end of this FAQ for a few mechanical problems that a TRV900 might develop.

I am not a vendor myself. If you wish to locate an authorized Sony vendor, call Sony. As of Feb. 2002, that web page lists USA phone numbers for Product inquiries and Dealer/Servicer locations as (800) 222-7669 (voice), (941) 768-7669 (voice), (941) 768-7790 (fax).

Cautionary Tales Department: some mail-order and online places advertise amazing rates on camcorders in the backs of photo magazines, on the Web, etc. but if you believe you'll actually be able to get these prices, I advise you read this note and the letters here. This camera is very popular, so many retail and mail-order outlets do have it, but I'll let you do your own shopping. Please note, if you're not familiar with a company it's good practice to check their rating before buying from them, with the Better Business Bureau. Reputation is important. You might search deja.com to see if any relevant consumer experiences have been posted to a newsgroup. Please consider the survey of mail-order photo/video stores, the reseller ratings site, the complaints at consumeraffairs.com and the information at photo.net before ordering from an unknown source. As user comments there document in horrifying detail, certain companies advertise great prices and then may do a number of things you didn't expect or discuss, once they have your credit card. For example, regarding CCI (Camera City Inc.) read note 61 and 71 on this page. Outright fraud or theft, while rare, can happen: for instance read this story. Don't let it happen to you.

Shameless Commerce Division: One Call will give me a 1% referral if (as a first-time customer) you buy via that link, or mention my name, John Beale (beale@best.com) on the phone. That is the only commercial arrangement I have related to my web page. Some vendors do not advertise the lowest prices, and apparently expect you to bargain. The comparison- shopping is up to you. When I'm buying a camera or accessory, I check out all the dealers I'm familiar with (in my case, generally the six mentioned above) who advertise the item, and buy the cheapest one that's in stock- no doubt you'll do the same, with your own preferred dealers. Some other places do advertise tempting rates, but remember to check their reputation along with their price.

Warranty: The standard Sony warranty in the USA is 1 year parts, 90 days labor. My Sony TRV900, TR7000 and VX2000 camcorders and MiniDV "walkman" deck all came with the same warranty, which is reproduced here. When I got the camera I got an additional 4-year warranty from Mack for $300 (it is now around $125 I think). This is the Mack agreement. They don't cover "professional use" and they require you to ship the camera to New Jersey and pay for shipping both ways (the card says ...or local service (at their option) but when I needed service and called, they said to ship it). If they determine they can't fix the problem, they will then ship the camera to Sony. I couldn't afford this compound delay so I brought it to the local Sony center and paid out-of-pocket. Sony charges a flat rate for most repairs, currently around $215. If I had to do it again I'd use Sony's extended warranty. The warranty is just an insurance policy; you have to decide if you want to pay now, or (possibly) pay later.

TRV900's marketed in Australia & New Zealand come with a "Domestic Consumer Products Service Seal" card in the box, specifying that video cameras used for "Domestic or Private Use" come with a 1 year parts & labour warranty, while those used for "Commercial or Professional Use" come with a labour warranty of 3 months & parts for 1 year. -South Caulfield

I asked SONY in Germany and they say that a SONY product has a world wide warranty of at least 6 month depending on the rules in the specific country the service is done. I asked then: What if I buy a SONY camera in the USA and have it serviced in Germany? The answer: Warranty time in Germany (by SONY) is one year full service. I have to present the bill that shows the date of purchase and service will be done. -Peter Plewka

For anyone buying a TRV900 or considering an extended warranty... I called up SONY USA and asked how much was their extended warranties if I purchased directly from SONY: 1 year: $99. 2 year: $195. 3 year: $274.95  I had bought my TRV900 at B+H photo/video Tel# 800-606-6969 (http://www.bhphotovideo.com/) for $1999.95. -Kwang Shieh, July 1999

[07/08/2002] Yes and no. On April 4 2002, Sony announced a new 3-CCD model, the TRV950 with HAD-type megapixel CCD chips, and that camera actually became available in the USA starting in July 2002. The new CCDs are slightly smaller than 1/4" and may not do well in dim light: they are rated at 7 lux in normal video mode (the TRV900 is rated at 4 lux). This page is a good technical description of how light sensitivity should be measured (but usually isn't).

The TRV900 was released in the US in August 1998 and sold until mid-2002. Many people, thinking about product lifetimes of computers and inexpensive consumer cameras, assume that nothing that old could compare with the latest model. People are still using the TRV900 because it is still a very good camera as compared with other compact 3-CCD video cameras now on the market. The various 1-CCD cameras have improved over past versions, and have higher-resolution stills, but most who've compared them directly still seem to prefer the TRV900 for video quality. See also this excellent comparison page, you can ignore the Japanese characters; just click on the model names for sample images. For example, the newer, smaller DCR-TRV30 is worth considering, but I'm told it does not perform as well in low light as the TRV900.

[5/18/2004] I used the TRV900 for four years and it always served me well. In 2002 I sold it, and I currently use the Sony DCR-VX2000 and Panasonic DVX100 MiniDV cameras for paid jobs; these are larger and more expensive tools (the current models are VX2100 and DVX100a.) The VX2k is especially good for event video with mediocre to poor lighting. However, for casual video the TRV900's smaller size is an advantage, and with enough light it can give you professional quality results.

You could look at the "minimum lux rating" advertised by video camera manufacturers, but this is widely regarded as useless; it's really a marketing fiction. This page is a good description of how it ought to be measured (but isn't, for US consumer products.) I have some low-light example stills showing color response here. Or, you might compare the effective ASA speed: in comparing the TRV900 with my Canon Elan II (35 mm film camera) I measured an effective speed rating of ASA 154 (normal scan) and ASA 65 (progressive scan). My tests here show the Canon Elura=ASA 77, TRV10=ASA 55, TR7000 Digital8=ASA 28. According to Richard Trahan (rtrahan at monmouth com) the XL1 is ASA 200.

Effective ASA rating means, that's the film speed you'd need to use in your 35mm camera for a properly exposed image using the same f/stop and shutter speed as the TRV900. These numbers are for 0 dB gain in the camera. In auto mode and dim light, after the iris opens up all the way, the camera will start to employ electronic gain up to a max of +18 dB. This gain will increase the effective film speed, as follows:

 Gain   Eff.speed   resulting video image
----------------------------------------------
  0 dB   150 ASA   excellent image quality
 +6 dB   300 ASA   still a pretty good image
+12 dB   600 ASA   rather grainy, flat colors
+18 dB  1200 ASA   very grainy, dim colors
----------------------------------------------
Obviously the "resulting video image" column is very subjective. In casual viewing you might not notice a +6dB gain situation but you probably would notice +12 and certainly +18, due to the inferior picture quality. Actually I could make similar comments about the print quality from 35mm film stock, too (relative to slow film that is- even high-speed 35mm film has more resolution than video).

Another way cameras are rated is the f-stop selected in auto mode for a given illumination (for instance the $10,000 Sony DSR-300 is rated at f/11 at 2000 lux). I have heard this is based on a 90% reflectance (rather than the 18% grey card used in film photography). My measurements indicate the TRV900 will operate at f/6.7 at 2000 lux for a 90% white target, or f/3.4 at 2000 lux for a uniform 18% grey (1/60 exposure, normal interlaced scan, 0 dB gain, auto-exposure). The TRV900 has a f/1.6 lens at the widest angle setting. So, you will need 440 lux for a bright video image with no graininess (for auto-exposure to set f/1.6 at 0 dB gain for a 18% grey subject). (Conventional photographic wisdom is that the average reflectance of the average subject is 18%; your subjects may vary.)

So what does this mean, you may ask yourself? Well, 440 lux = EV 7.5, which the very useful exposure computer subjectively describes as somewhere between stage show and indoor sports, and night football lighting. If you use f/1.6 plus 6 dB gain, you have a reasonable but not ideal video image at 200 lux (EV 6.5), still somewhat more than a home interior at night. I measured that level at a distance of just one foot from the lampshade of a 100W light, or 4 feet away from the bare bulb. (You can see why professionals use bright video lights!) At the low-light extreme, if you use f/1.6 and +18 dB gain (8x more light sensitivity) and 1/4 sec exposure (15x more sensitive than 1/60 sec) I calculate 3.7 lux is needed for this very jerky, grainy video image... but maybe this is where Sony gets the 4-lux sensitivity they advertise for this camera! Remember if you zoom in, the lens is f/2.8 on the long end, which means 3x less light enters relative to the widest setting. By the way 1 foot-candle = 10.74 lux. (Note: an earlier version of this FAQ had an incorrect Lux<->EV conversion. I believe this one is correct.)

Subjectively, the camera can see about as well as I can with my naked eye at 1/4 sec shutter with no additional gain. It can see more than I can when you add +18db gain but again the signal is rather grainy/noisy. Gain is also manually controllable from -3db to +18 in 3db steps. See also my collection of comments from email/news here.

By the way, if you don't need color, the inexpensive black-and-white "security" type cameras are better at low-light than any color camera you can probably afford. One reason is that they don't have the infrared filter that color cameras need to reproduce good color, so they make use of a wider spectrum of photons. I believe the "night shot" mode on some Sony models (not the TRV900) just mechanically removes the IR filter, essentially turning it into a B/W security camera. Nightshot models also have a small built-in IR illuminator (IR LED's).

Uninteresting Technical Details: Bare CCD chips respond to the full range of visible light (plus near-IR), so a plain CCD chip gives you a B/W camera. If you want a color image, there are two options: filters or a dichroic prism. A single-chip CCD camera has a colored dot over each pixel, much the way your TV is composed of small clusters of red, green, and blue dots. These dots act as color separation filters which absorb part of the light: for instance, a "red" CCD pixel has a filter over it that absorbs green and blue light. Some models have primary (RGB) filters, some use complementary (cyan-magenta-yellow) colors. In any case, much of the incoming light is absorbed in the filter and goes to waste.

The dichroic prism in a 3-chip camera simply divides the incoming light into red, green, and blue components and routes each part to its own chip-essentially all of the light (I'll guess better than 90%) actually gets to a CCD chip surface where it is useful in forming an image. (Note that this prism is a beamsplitter with internal interference-type (reflective, non-absorbing) filters, and not a dispersive prism of the sort you can make a rainbow with.) There are additional complications since CCD chips have a dark level which varies with temperature and individual pixel, and there is electronic readout noise; both these factors make spreading your photons over three chips more noisy than using a single chip, other things being equal. Different CCD chips have different fill factors, surface reflectivity, and quantum efficiency. In sum I think the 1-chip vs. 3-chip low-light debate to be difficult to decide "a priori"- you need to compare the actual cameras directly and choose on that basis.

With the TRV900, as you increase the video gain from 0dB (or more commonly, you use an automatic mode and the camera adjusts it) you will see the colors become less saturated and the video noise or graininess of the image increase in proportion. Using a slow shutter speed (down to 1/4 sec) doesn't add noise or subtract color, but it's harder to focus (lag time before response) and gives you a "stroboscopic" look to motion. For still pictures of still objects, from a tripod, it works fine. If you can get it in focus :-).

NOTE: If you use progressive mode in dim light, all other things being equal, you will notice the image is one full stop darker than in normal-video mode. It's not clear to me why this should happen, but it does. You'll need more light, or use extra aperture, gain, or lower shutter speed to compensate. You can't go slower than 1/60 sec. in "memory" (PC Card/disk) mode. You can go down to 1/4 sec exposure while in video (tape) mode, but you might as well use interlaced scan in that case. To my eye, the actual progressive mode resolution isn't any different from interlaced mode when used below 1/60 shutter speed.

Note that while the filter diameter on the TRV900 is 52mm, the actual final lens has a diameter of 27 mm, which is similar to cheaper compact camcorders. One advantage to a larger filter diameter is less chance of vignetting with one or more filters stacked together. When I use my Kenko 2x teleconverter, which has a 67mm objective, I seem to get a brighter picture (for equivalent total focal length), because I can use the internal zoom at setting with a wider f/stop.

[01/01/01] People who shoot weddings usually need the best possible low-light performance. One wedding videographer suggested that an older, 2/3" single-chip camera would work better in low light. This makes some sense; it would have 2.4 times as much total CCD area as three 1/4" chips, and presumably a correspondingly larger lens to let the light in. By the way, if you have $2500 for a wedding camcorder I would recommend the Sony VX2000, due to its superior low-light performance. Remember, both TRV900 and VX2k cameras can use slow-shutter for improved low-light response. This makes motion appear more jerky or stroboscopic, but 1/30 sec shutter is usually ok, and 1/15 sec may be ok for scenes with very little motion. 1/8 and 1/4 sec is more of a special effect, and of more limited use, but try it out sometime and see! You may also find these wedding tips useful.

XL1 is somewhat better. Mark Grant Tue, 24 Aug 1999

In direct comparisons on the feature shoot I'm working on the XL1 is somewhat better in low light than my TRV900; it's not dramatically better, but there's definitely less noise in the shadows on the XL1 when using identical exposure settings (...) And by low-light we're talking rooms lit by burning torches here, not properly lit sets.

TRV900 is ok for weddings. Bob Andersen Tue, 24 Aug 1999

I use a TRV900 and an XL1 as my main cameras and a TRV9 for my cut away camera:

1. In almost all shots the TRV900 looks as good as the XL1. It certainly looks good enough.
2. I use the TRV900 and the XL1 for receptions.

-in good light the xl1 & the TRV900 both look great
-in low light they both look OK with small on camera lights (20-30 watt)
-in very dark the XL1 looks better than the TRV900, but the V900 is still good enough.

I charge between 1500-2000 per wedding & will be raising my prices to 1800-3000 depending on the options (In Raleigh, NC). I edit them on a NewTek Flyer. I've done 30+ weddings over the last 3-4 years. What I have found is:

-you have to control the expectations of your client
-whether a bride is happy or not happy is based on the content of the video and not the image quality.

When I show the bride & groom their video there are always some parts that are too dark or a little too grainy because I boosted the brightness with a proc amp. You know what? The B & G never notice. They will laugh because Aunt Tessie is doing the twist, and cry if 90 year old grandma leaves her wheel chair to dance, BUT WILL ALMOST NEVER complain about the video quality. It's a content thing pure and simple.
-A customer will notice bad audio right away.

IMHO the TRV900 is perfect for weddings. If I had to do it over again I would have bought 2 TRV900's instead of the XL1. Please note that I would have to verify that the V900 could handle my audio needs...the XL1 does a great job here.

No. The TRV900 is a 3-CCD camera which is optimized for the best color video. The CCDs are permanently bonded to the dichroic beamsplitter which determines the color response (and blocks most infrared). There are no 3-CCD type cameras yet available with "nightshot" capability.

Most Sony single-CCD cameras have a "nightshot" mode, which gives better performance in dark conditions by physically removing the IR-cut filter in front of the CCD (used to improve color imaging in normal light), along with a built-in infrared LED to illuminate nearby objects. It is a monochromatic mode, and somewhat unnatural-looking due to the differing reflectivities of objects in the near-IR (700-900 nm wavelengths). If you want to take images of nocturnal animals, do surveillance, or any other video in very dark conditions, you need one of the single-CCD cameras that does have nightshot.

This is a very frequently asked question, or variations involving these or other cameras. Of them all I have used only the TRV900 and GL1, both of which I'm happy with. All I can say is: in good light, 3-chip cameras always have a better picture than one-chip cameras, and they are almost always more expensive. In good light, the difference between a good 1-CCD camera and a 3-CCD model may be hard to see, especially on a consumer TV screen. Each camera has different features, capabilities, and price. I have some reviews of current models here, and I have some comments from others comparing the TRV900 with many different newer and older models here. See also the various reviews by video magazines. The PD100 is the "professional" version of the TRV900 and is sold by Sony Professional (if you call the Sony consumer division, they won't have a listing for it). The current model (Fall 1999) is the PD100a, a "fixed" version which avoids the hollow or ringing sound some people observed in the PD100 audio. The PD100/PD100a is a different color, but is mostly the same inside and out as the TRV900, with some exceptions as noted below. I have a feature comparison (of the PAL versions) from Sony Professional. Vincent Chan has some side-by-side photos from the two cameras (PAL version). The information below based on a post to rec.video.desktop by (malcolm at abbeyvision demon co uk).
Differences between Sony DSR-TRV900 and PD100:

The TRV900 records on 60 minute MiniDV tapes in SP mode (1 hour) or LP (1.5 hour). The PD100 records in DVCAM format only, which is 1.5x faster than SP mode. Thus, you get 2/3 the recording time (40 minutes instead of 60 minutes per tape). MiniDV and DVCAM formats store identical audio and video information, but in DVCAM the audio samples are locked to the video timebase, and (due to wider track pitch) has better interchangability between decks (Sony decks of course, since DVCAM is a proprietary Sony format). The compact-format 40-minute DVCAM tape is physically the same size and length as the 60-minute MiniDV tape, and you can also use tape labelled DVCAM in a TRV900, likewise MiniDV tape in a PD100. I don't know if there is any physical difference between the tapes, apart from the label (and the price). Apparently, both cameras can read both video formats, but only record in their "own" format.

The PD100 (DVCAM) has audio locked to the video sampling rate, and the TRV900 uses unlocked audio. Since MiniDV is only unlocked short-term (it maintains at most 1/3 frame offset longer term) there may be little difference in many applications. See www.adamwilt.com/DV.html for the details and the implications.

The PD100 claims 12x zoom on the side. The TRV900 claims 48x. In fact, both have 12x optical zoom and 48x digital. (Like all digital zooms, it looses resolution and I don't find it of much value.)

The PD100 has two zebra settings (IRE70, IRE100). The TRV900 has only IRE100.

The TRV900 has "LaserLink": IR composite connection to a TV, the PD100 doesn't.

The PD100 allows timecode reset to zero. Not a lot of use but it's missing from the TRV900.

Internally the PD100 has (according to Sony) handpicked components. Sony claim the CCD block is noticeably better. Also that the PD100 has a stronger case(?). I saw a Sony demo where it was shown (not all that clearly) that the weaker TRV900 can flex under the weight of heavy lens attachments and the PD100 doesn't.

The PD100 comes with a wideangle adaptor with hood, and a XLR adaptor which can provide +48v phantom power.

In Q1 1999, Sony released some cameras which store DV format video on existing Hi8 format videotape, and in Q2 2000 released "second generation" models using the same CCD chips. Hitachi (Q3 2000) also has a few models. They call this new format Digital 8. Since it is a standard DV datastream, the recorded video has the usual DV specs like 500 lines of resolution and much better color than Hi8- but that's just the recorder, not the camera itself. For all the current Digital8 cameras, reading the fine print you'll see the camera does not achieve 500 lines.
2. Maximum resolution only when recorded through the video inputs. -footnote, Sony Digital8 web site
By contrast, independent tests show the TRV900 camera unit really has 530 lines of resolution live, and 500 lines in playback. There is a fairly clear dividing line between 3-chip and 1-chip cameras, both in price and image quality. The former are pro or semi-professional devices and the latter are more consumer. For instance the Sony TRV-9 is a 1-chip DV which is considerably less expensive than the TRV900, or the Canon Optura, etc. If you don't need the image quality of a 3-chip device, there is a wide range to choose from, both in MiniDV format and Sony's new Digital 8.

The D8 cameras are the cheapest digital video recorders now available. I have the cheapest, the TR7000 and use it as a digital VCR. They record true DV signals over firewire just like any other MiniDV or DVCAM recorder. I put some more notes and a few still frames here. If you don't need the still-image features and 3-CCD image quality of the TRV900, a Digital8 model might be an attractive choice.

Canon announced the GL1 in July 1999, and it actually showed up in stores in October '99. As of Oct. 26 1999, the mail-order price is around $2300. It seems designed to compete directly with the TRV900 (unlike the Canon XL1 model, which is much larger, and more expensive). I purchased a GL1 in December '99 and made my own GL1 web page with comments and comparisons.

[05/18/2004] The current model is the GL2, you can read about that camera at the Canon GL2 Watchdog.

Sound Quality, Noise

I think the built-in mic sound quality is fine, and it usually doesn't pick up any camera motor sound (see below). The built-in mic does not have as much bass response or as clearly separated a stereo image as the external stereo mics I've tried (Sony ECM-MS908c and Audio Technica AT822, see mics page).

The camera has a stereo mic input jack ('plug-in power': it supplies about 2.3V bias through 10k ohms, for powering electret mics). From a menu button you can have a sound level meter show up in the viewfinder, and adjust the record level (or set it to AUTO). The meter and level adjustment is mono, that is acting on combined L+R audio. You can select 16 bit at 48kHz, or 12 bit at 32 kHz audio sampling, but there is little reason to choose 12 bit audio.

If you do record audio in 12 bit mode at the SP tape speed, you have the option to dub a second audio track later, after recording the video. Unlike the Canon XL1 you can't record two stereo tracks at the same time, you have to rewind and add the second one as an audio dub later. You can dub second-track audio from the built-in mic, the mic jack, or via the A/V input jack (which doubles as output jack). You can't hear the first track while dubbing the second. This capability to dub a second track is very seldom used because computer-based video editing (NLE) is now commonplace, and it offers much more flexibility and convenience.

Note, the A/V line level input is active only in VCR mode. In camera mode (recording video through the lens) you can only use the mic jack for external audio signals. If you have an external line-level source, you need an attenuator (some XLR adaptors will do this.)

No, which is very strange. When the camera is recording, I can hear a definite noise from the tape drive, I'd say it's a buzzing-type sound which I can hear from up to a foot away in a quiet room. Remarkably, this noise does NOT come out on the soundtrack. I used the internal microphone both with AUTO gain control, and manual control with the gain all the way up, in a quiet room. I could hear no trace of the tape transport sound on the soundtrack when I played it back. Yet the microphone does pick up other sound well. I can only assume the microphone is directional and sound from the rear of the microphone, where the camera is, is not picked up.

I just tried the experiment in a closet to muffle even the faint outside traffic sounds. My voice was deafeningly loud on the recording, but when I was quiet, I could hear no motor noise. Correction: there was a little motor buzz-noise when I held the recorder near the door (a large, flat, hard surface). My theory is the sound coming off the tape drive bounced off the door and back to the microphone. How it avoids getting conducted directly through the case I don't know- but it works!

Another interesting note: at max gain, if you are listening as you record on headphones, you will hear a quiet buzz when the camera is in standby. This goes away, entirely, when you start recording- even though just listening with your ear next to the cassette compartment, the buzz from the tape drive is louder while recording. Mysterious. At max gain, by the way, you can rub your fingers gently together several feet from the microphone and it sounds loud.

Later note: Well, I can hear some buzz-noise on the soundtrack during playback at the moment the camera starts recording, but it fades quickly to inaudible over about one second. It might be a DSP (digital signal processor) with an adaptive filter algorithm which is "locking-in" to the interfering signal and cancelling it.

I tried recording myself playing my classical guitar, which is a quiet instrument and difficult to record under any circumstance. The TRV900 picked it up fairly well, along with the refrigerator in my apartment when that came on. There was some hiss which you get with any small condenser microphone set at high gain. Note this was only an experiment- as with any camera, for critical recordings, I'd recommend a suitably located external microphone. I describe a few such mics here including sample sound clips. I think the TRV900 is fully adequate for home recording. If you have truly professional audio requirements you may need to resort to an external MD or DAT, then align the soundtrack with the video in post-production. See also the microphone, XLR adaptor and audio mixer sections of my accessories page. Note, there is apparently some unit-to-unit variation in the noise level of the camera's mic preamp.

Subject: Re: Hiss on the TRV900?
From: "Clifand" (clifand at aol com) 
Date: Wed Apr 12 14:33:05 2000
Newsgroups: rec.video.production

> No annoying levels of hiss on my TRV900. I monitor recordings with
> headphones (post production) hiss is not a problem. I believe the TRV900
> is nearly as good as my Sony DAT recorder

My first TRV900 had an unacceptable level of hiss.  I returned it a got a
second one without the hiss problem.  At that time, another participant in
the newsgroup said he was aware of a hiss problem but obtained a TRV900
without it. For over a year and until now, I have not heard anything about
this hiss problem.  -Clif
I have tried recording with the camera both ways, with the AC adapter and without. I used AGC in a quiet room (which turns the gain all the way up). I listened to the playback with headphones straight from the camera, with the volume turned all the way up. I could hear no (interference-type) noise either way. I did hear the ticking of the clock, my footsteps, breathing, distant conversations, etc. The zoom lens makes a barely detectable sound, but you don't hear it at all unless you're listening carefully and using this unrealistic max-volume setting. The TRV900 has a stereo minijack for an external stereo mic. The left and right channels are separate so you could use two separate microphones. The most convenient way is to use a box like the Studio1 Productions XLR-PRO (see accessories page), which mounts under the camera and gives you two separate mic connectors (mic or line level) with independent volume knobs. If your mics have miniplug connectors and you don't need separate volume controls, a much cheaper option is a stereo y-adaptor like the Radio Shack 274-375B connector. Note that miniplugs are notoriously fragile and don't always make good contact. Make sure the adaptor, and each mic plug goes all the way into the socket.

If you want to record sound in two places, you must use two external mics; you cannot record using both the camera's internal mics and an external mic at the same time.

Yes, you can connect to an external mixer, mic preamp, or other line-level source. The usual way is to use an XLR adaptor such as those made by Studio1 or Beachteck. See also the Accessories:XLR Adaptors page. You connect a cable from the external line-level source into the XLR socket on the adaptor, and in turn connect the adaptor's stereo miniplug into the TRV900's mic input. These boxes have their own volume control. Note that they are switchable between "MIC" and "LINE" level, make sure to select "LINE" if in fact you are getting a line-level feed.

The stereo minijack TRV900 mic input features so-called "plug-in power" which means the camera injects a few volts of bias on the mic input, which can power some small consumer-type electret mics that don't have their own battery (although most do have batteries anyway). This is a feature common on Sony cameras and recorders. It is NOT the same as, and not to be confused with the +12 or +48 V "Phantom" power which is used for powering professional condensor-type mics on balanced lines (typically with XLR connectors). In any event the XLR adaptors referenced above use transformer coupling, so the TRV900's mic input DC bias does not reach the external source and does not affect it.

Still Photos

There are several different ways to get still photos onto a floppy/PCMCIA flash card/memory stick. One is to select "memory" mode and press the photo button (either on the camera, or on the remote). This takes a progressive scan image from the lens directly to the memory device and you do not even need a tape in the camera. Note that pressing the photo button halfway down freezes the image, but you must continue to press the button all the way down for the red bar-graph to appear and the image to actually be stored on the memory device (flash card/memory stick/floppy disk). Another way to get stills is to select "camera" mode, take a normal video on tape, and then play the tape back in "VTR" mode. You can pause the movie at any time and press the photo button, which grabs the frame from tape to the floppy/PCMCIA. If you press the photo button while in camera mode, you will get seven seconds of the still image recorded on tape, not on the floppy. Later when you play back, you can also transfer these still images from tape to floppy, just like frames from the moving video. You have to switch to progressive mode yourself if you want the better resolution when you hit the photo button, IF you are recording your still photo to tape. Recall that the camera has four modes: Off, VTR (tape playback) Camera (tape record) and Memory (record to/from PC Card). If you are in "memory" mode, you get progressive mode automatically, no choice. So if you record from the lens direct to your PC card, you'll get the higher resolution progressive scan all the time, but recording to tape gives you either progressive or interlaced, whichever mode is set. Just to complicate matters, in VTR mode you can also capture a still to the memory device from external S-Video in, composite video in, or DV (firewire) in. In these cases the still is always saved in interlaced mode. One person wrote me to say they use their TRV900 almost more as still than a video camera. Before I had one myself, I would have been surprised to hear this, but it makes sense to me now- this camera really blurs the line between still and video. I have a Canon EOS Elan II (decent 35mm film camera) which I've used a lot in the past year, and also a Kodak DC120 "megapixel" digital still camera. Both of those have higher resolution (the 35mm especially) and film will give you greater dynamic range, but the TRV900 can turn in some very attractive still shots. Judge for yourself here. The "progressive scan" mode makes it higher resolution than a normal video capture. The 12x zoom lens is not that impressive for a camcorder, but is very long for a still camera. You can store your still as regular video on the DV tape, or on the "memory device". The camera has a PCMCIA (PC Card) slot, and comes standard with a 3.5" floppy drive (PC format) with a PCMCIA adapter. I'm told this drive also works with a laptop (under Windows 98) if you download the driver. The floppy is handy for transferring a few pictures to your computer in JPEG format, and since it goes the other way too, it's great for adding fancy (still) titles to your videos without having a full NLE setup. You can also use a compact flash card or Memory Stick with appropriate adaptor, up to 128 meg in size. You can use one of three image qualities when storing stills:

Name         Compression   Approx. size
----------------------------------------
SUPER FINE   1/4           150k
FINE         1/6           100k
STANDARD     1/10          60k
----------------------------------------

All three types of JPEG image files have 640x480 pixels, so your images will have the proper 4:3 video aspect ratio when viewed on a standard computer monitor or printed. The DV format video is 720x480 with non-square pixels (for NTSC cameras anyway).

The camera uses PCMCIA Type II ATA flash cards. (The only difference between Type I, II, III PC-Cards is thickness. Type I is thinner and ought to work but I haven't tried it. Type III is thicker and won't fit.) They are sold by many computer stores for instance www.buy.com or alix.com. Full-sized PCMCIA flash cards are rare and expensive, so what everyone uses is either a "compact flash" card, or a (Sony proprietary) "memory stick" in a PCMCIA adaptor. I have a 30 MB SanDisk and a 128 MB Mr. Flash compact flash card, and both work fine. They go into a full-size PC Card (PCMCIA card) adaptor (about $10, sold wherever flash memory is sold) which in turn goes into the camera. The camera indicates I can store 152 images in "super fine" quality. This varies depending on the detail in the images: after taking a few snapshots the camera revised this estimate to 153, then 154. There is something called "SmartMedia" which can fit in a type II PC-Card adapter. One user reports this does not work correctly with the TRV900. Another said it works fine. The one reported to work is Philips SmartMedia memory card AY3957 (3.3v 8M), along with the Philips SmartMedia Adapter AY3951 (works with 3.3V and 5V memory modules) If you're buying a flash card, also check my accessories page.

You can transfer still images from the PC Card to the DV tape, or the other way. There is a "slideshow" mode in which the camera will display each of your images in turn from the PC Card, for two seconds. (Unfortunately you can't transfer the whole slideshow to DV tape at once, unless you have another, separate DV recording device.) If you're taping an event, you can wander around a bit before and after the event taking still photos onto the PC Card, which you can later import into your PC, superimpose titles on, and then write back to the tape at the appropriate points. Voila, excellent digital quality titles, without a Firewire card! See also "How can I play back JPEG files" below.

I bought a DataChute PCMCIA Card reader (Antec Inc. Fremont, CA http://www.antec-inc.com ) for my desktop PC and it sometimes worked on my home system. Later I got an Antec USB PhotoChute PCMCIA/SmartMedia reader and it worked fine until I tried to do video editing when it caused glitches. I currently use a Comotron USB 6-in-1 reader (compact flash, memory stick, Smart Media, MMC, etc) which works fine all the time under Win2k.

Normally the camera will always start over the filenames from MVC00001 when you erase the memory card. But if for instance you leave a "dummy" file named MVC00029.JPG on the card, the next photo stored will be named MVC00030.JPG, and so forth. In this way you can make all your photos have unique filenames. The filenames will wrap around after MVC09999.JPG and after that the camera will not display the images in playback mode, although they are still safely stored on the memory card. (thanks to Matthias Kasimir for pointing this out... that is a lot more stills than I've taken!) There is a small slide switch labelled "Memory Release" on the bottom surface of the camera near the tripod mount point. This is connected to an internal mechanical lever which ejects the PCMCIA (PC Card) flash card from the camera's memory slot, if any is inserted. Not that I know of. I have the small Sony video flash, and also a variety of photographic flash and strobe units used with my 35mm camera. You can use the small Sony video flash to trigger a larger flash or strobe using an inexpensive photo slave device. (A "photo slave" is a small box, about $25, with a photosensor and a PC cord jack: it sees the small flash and closes the contact on the PC connector to trigger your large one. There is a delay between the trigger and main flash, but it's small; in my case I measured the delay between flashes to be 850 microseconds).

I've used this setup to check lighting ratios in a 3-point strobe lighting setup, and it worked pretty well. In this way I am using the TRV900 to replace a polaroid camera which you'd normally use to do the same thing. The '900 does lack the dynamic range and resolution of film, but of course it's instant results, plus zero incremental cost per shot.

You can adjust the camera's exposure by changing the f/stop, and with an external flash you'll probably need to. You can also change the shutter speed but you loose flash synch any higher than 1/90 sec, giving you a dark frame.

You can modify still frames, and add titles and other graphics on your computer and write them back to videotape, using the floppy disk or memory card. This gives you great flexibility without the expense of a firewire card, but there are a few tricks to keep in mind. First, remember that TV sets have "overscan", meaning about 10% of the image area around the edge of the JPEG picture will not be visible on the TV screen. Each TV is slightly different, you may have to experiment a little. Second, you must use the exact file name format eg: Mvc-0001.jpg and use 640x480 resolution, and do not use interlaced JPEG format.

Specifically: I've had success with modified TRV900 pictures and also scanned images from other sources. I've tried Adobe Photoshop 4.0.1 LE, ULEAD PhotoImpact SE 3.01, and ThumbsPlus 3.30 (shareware from www.cerious.com). Photoshop is the most difficult. Here's what to do:

1) Make sure the image is "True Color" (not greyscale, 256 colors etc.)
2) Scale your image to exactly 640x480 pixels in size, if it isn't already.
3) Make sure you are NOT using jpeg "progressive compression" format.
4) Do not save any image comments, thumbnail previews, etc.
5) Use a filename like MVC00001.JPG for a PC Card or MVC-0001.JPG for a floppy disk.
In Photoshop, turn off image previews: choose File>Preferences>Saving Files and select "NEVER SAVE" in the "Image Previews" box. In JPEG Format Options, choose "Baseline ('Standard'). When you save the file, confirm that the "Save Thumbnail" box is NOT checked. The camera can read Photoshop images with JPEG Quality 1-3, but does NOT work with Quality 5 through 10, I don't know why.

The camera can read files saved from PhotoImpact or ThumbsPlus in any Quality setting from the minimum right up to 100%. However, I have looked closely at the video output with highly detailed test files, and can see no differences above Quality=85%.

None of the programs I know of will generate a valid "index" image for the TRV900 6-photo-per-screen index mode, they just appear as black squares; but the full-screen version of the image should work. All of the above is on a PC/Windows platform. I've been told that for writing JPEG images that the TRV900 can read using the Mac, Photoshop doesn't work, but GraphicConverter from Lemke Software does. 

Briefly, televisions and computer monitors have different "gamma" values, which the camera software does not correct for. Gamma and gamma correction is complicated enough to have its own FAQ on the web. In addition, JPEGs from the camera often do not use the full range of intensities from 0 to 255 (check the image histogram in a processing program and you'll see). In short, I'll just say that yes, the pictures look darker when imported as JPEG files on the floppy into your computer. The pictures on my home page which weren't manually corrected, did look brighter on the LCD monitor and on the TV screen.

You can fix this manually in your paint program by using a gamma correction factor of about 1.2 or so. You may want to slightly increase brightness and contrast also. For instant results, if you are using Photoshop, you can try Image->Adjust->Auto Levels (Shift+Ctrl-L) which often fixes things in one step. If Photoshop is too expensive (or even if it isn't), I recommend the "ThumbsPlus" shareware image utility from www.cerious.com-- I use both programs frequently. "Thumbs" has several auto-processing functions, for instance Image->Process->Improve Video Capture (Shift-V) which does a histogram stretch, plus a "sharpen" step. If you are adding titles and then sending them back to the camera, remember that any "gamma correction" you do will probably make the image too bright on the TV! I've tried to adjust my computer monitor brightness and contrast settings to match the TV images, with some success.

Here is an email from Ira Solomon:

Opening a TRV-900 JPEG in PhotoShop 5 was terrible. Dark doesn't begin to describe the result. Almost black. And I found it almost uncorrectable. Intellihance produced a poor result, although better than the original. I eventually found out the cause. The fix requires going to Preferences on the File menu. Under that you want profile. In there change the mismatch setting to "ignore". Now the pictures will open in a usable way. This problem will not occur in the previous versions of Photoshop.
All JPEG files (written to floppy, flash card, or memory stick) are 640x480 resolution. It does not matter if they're progressive or interlaced, or normal, fine, or superfine quality. Those things determine how sharp the actual image looks, but they're still all exactly 640x480. The JPEGs from interlaced video are fuzzier because the camera interpolates (invents) half the pixels based on a single video field, to give you the 640x480 image. The progressive scan mode does not invent any pixels, giving you a sharper image. If you have a firewire card you get DV data from the camera which is (always *) 720x480, rectangular (non-square) pixels. 99.99% of all computer monitors use square pixels so you'll have to readjust it to 640x480 for the aspect ratio to be right anyway. Supposedly there are fewer compression artifacts in DV as compared with JPEG but I don't see a big difference. They're both quite good. Dave Millman has a comparison of JPEG vs. firewire stills here. There is essentially no way** to get more resolution than 640x480 JPEG or 720x480 DV out of this camera- for that you want a HDTV camera, or a digital still camera, or better yet a film camera. However if your subject is not moving (eg., a landscape) you can take several shots and combine them with software; see below. Some people find the "Genuine Fractals" scaling software from Altamira to be useful in increasing apparent resolution. Chris Reijnen scanned in a photo print made from a fractal-rescaled TRV900 image here which I was impressed by.
* unless you use a PAL unit, which is 720x576 through firewire, but still 640x480 for JPEG stills.
** unless you digitize the analog video live, and oversample it, which somewhat improves resolution, see here.
By taking many different stills with some overlap between them, you can make very wide combined images. This can be done in digital still mode or even progressive-scan video with shots selected after the fact (pan slowly or you'll have blurs, unless you use a fast shutter speed). To see what TRV900 panoramas can be like, see Peter McLennan's page.

There are several programs which make panoramas from two or more still images. Spin Panorama from pictureworks.com, PhotoVista from livepicture.com, and QuickStitch from enroute.com are three that various TRV900 owners have found to work. PixAround does interactive 360 panoramas for the web (a la Quicktime) using JPEG and a java applet. James Rigg's Panoguide.com reviews 14 software packages, and has everything you need to know about panorama techniques, and more.

You cannot get a truly clean, near-horizontal, thin (1 pixel wide) line on a computer, period. Either you fuzz out the line to occupy several pixels of width, or you leave it sharp and get stairsteps (aka "jaggies" or "aliasing"). If you have high enough resolution (enough pixels, that is), a fuzzy or antialiased line several pixels wide will still appear sharp, unless you greatly enlarge the image. This is an advantage of "megapixel" digital still cameras over the TRV900.

Still frames from the TRV900 have only 480 lines vertically at best. That's in progressive scan mode. In normal video mode, a still image uses only one field of a full frame (that is, half the 480 pixel vertical resolution) and that hurts you even more with near-horizontal lines. The exported JPEG is in fact 640x480 in either mode, but in regular interlaced video mode, every other horizontal scanline of the still is interpolated and the jaggies, for the most part, remain. If you want to take clean images with thin, high-contrast diagonal lines, try progressive scan mode (or better yet, use a film camera :-). Having said that, I feel the camera does quite well, for example this image (TRV900, progressive scan) has many near-horizontal lines and I see no jaggies at all.

The answer is both yes and no, depending how you mean the question. Anything displayed on the LCD screen also goes out the firewire port as video. So yes, you can use your NLE software and capture "video" while displaying your memory stick still in the camera, and then extract a still frame from it (of course all the frames will be the same while playing back a given still image). Many firewire capture programs also permit direct capture of a still from a video stream.

But no, you can't address the memory device directly through the PC's operating system via firewire as if it were a hard drive, it doesn't work that way. The camera's firewire port is limited to the DV protocol which permits only the standard DV (video) datastream to be transferred.

NTSC vs PAL

PAL and NTSC are two different analog broadcast video standards used in different parts of the world. The PAL video standard permits somewhat greater resolution and more consistent color reproduction than NTSC, and it runs at a slower frame rate. The TRV900 comes in two varieties: those sold in the USA and Canada record and play back using the NTSC video format, which is standard in those countries. Those sold most other places, including Europe and Australia, use the PAL format. The PAL version of the camera is properly called the DCR-TRV900E ("E" for Europe, perhaps) although many people still refer to it as the TRV900 for short. The NTSC version records video using 480 lines of 720 pixels at 60 fields (30 interlaced frames) per second. The PAL version records 576 lines of 720 pixels at 50 fields (25 interlaced frames) per second. France and Russia actually use a third video standard called SECAM but, as far as I know, there is no TRV900 model which uses that format. Johnson Ching reports that you can freely read and write PAL or NTSC tapes on either a TRV900 (NTSC) or TRV900e (PAL) camera through the firewire. Here is a letter from Jerome Maro explaining the situation:
What I understand about how those cameras are working is the following (tested with a PAL TRV900 and also a PC-1 (also PC-100, tnks James)).

1: PAL is 25 pictures a second.
2: NTSC is 30 pictures a second. (the number of scan lines also differs)
3: DV is neither PAL or NTSC, but comes in a 25 p/s and 30 p/s version.
4: Through the lens, a PAL TRV-900 records 25 p/s, a NTSC model 30.
5: From the digital port a TRV-900 (any model) records what comes in (25 or 30).
6: From the analog plugs, a TRV-900 will only record the standard it advertises (PAL or NTSC).
7: If you have a DV cassette recorded at 25 p/s (from a PAL camcorder...), and play it on a US TRV-900, that machine will output a non-standard signal at 25 p/s with NTSC 3.58 color encoding. Most TV will play it fine, but VCRs won't record it. From that tape, it will also output a standard PAL DV stream via firewire.
7bis: If you have a DV cassette recorded at 30 p/s (from a NTSC camcorder...), and play it on a PAL TRV-900, that machine will output a non-standard signal at 30 p/s with either NTSC 4.43 color encoding or PAL color encoding (choice in the menus). Most TV will play it fine, but VCRs won't record it. From that tape, it will also output a standard NTSC DV stream via firewire.
8: None of this cameras will convert the frame rate (or the number of scan lines). This is the key in understanding what the cameras do when faced with the "other" standard (that and understanding that you need different oscillators for 3.58 (NTSC) and 4.43 (PAL and NTSC 4.43) and the camera only has one).
9: The TRV900 will record still images from the PC card port in its native format only, i.e. if I put a NTSC-recorded cassette in it and add a few seconds of video from a jpeg file, that part is recorded in PAL.

If you are recording or playing back through the firewire (AKA i.Link, or IEEE-1394) cable, you are doing a "raw" digital transfer and there is no format conversion. If the original was PAL, the copy will be PAL, and the same with NTSC, regardless of the camera model (TRV900 or TRV900E).

If you have an analog video signal in one format and want to change it to another format, you need a video format convertor box. TENLAB is one manufacturer, and you can read more about them here. If you have a DV file on your computer you want to convert digitally, you need a software format convertor, for instance the one from Canopus.

If your only need is an occasional VHS tape in the other format (eg. NTSC to PAL or the reverse) to send to friend/family abroad, your best bet is simply to have it done by a local photo/video shop. In your local Yellow Pages under "Video Tape Duplicating & Transfer" you will find such services. I've done this and obtained a "first-generation" PAL dub from my NTSC MiniDV tape, by physically carrying in my MiniDV deck (or camera) to the duplication shop and having them take the A/V out cable directly into their converter.

Subject: PAL playback
From: Chris Williams (paint_progs at hotmail.com)
Date: 4/29/01
The most interesting aspect of the TRV-900 for me has been the PAL compatibility. But the 900 has been a little high for my budget. So I wanted to test to see if this ability was included in other Sony camcorders. So with the help of a friendly dealer, I spent the afternoon testing with a PAL DV tape.

No [USA model NTSC] Canon camcorders, including the XL-1, have the ability to play PAL DV tapes. [Canon does sell PAL camcorders (eg. XM-1) in Europe and elsewhere, but they don't play NTSC tapes. -jpb]

*All* the Sony Camcorders I tested were able to play my PAL DV tape, show the image on their LCD display and output to FireWire.

The camcorders I tested were:

   DCR-TRV900
   DCR-PC100
   DCR-PC110
   DCR-TRV20
   DCR-PC5
Nobody who needs a DCR-TRV900 should avoid buying one. But if all you need it for is to playback PAL DV, the dirt-cheap DCR-TRV20 (currently being sold at this local store for $999) might be a better choice.
Not that I'm aware of. According to my service manual, the PAL model has physically different CCD chips from NTSC (ICX216AL-13 TRV900, ICX217AL-13 TRV900E). By the way, it seems the red, green, and blue CCD chips are identical; it's their position on the dichroic prism beamsplitter than separates the colors. The PAL version has 450,000 pixels per CCD chip, and the NTSC version has 380,000. The frame rate (25 fps vs. 30) is also different. But if anyone manages to flip some magic switch and record PAL format on a 900 or NTSC on a 900E, please be sure to let me know. PAL has somewhat better resolution than NTSC, and also lower frame-rate (25 frames per second instead of 30 frames). PAL cameras are favored for transferring to 35mm film (which runs at 24 fps). I believe for casual use either video standard is fine. If you are going to use your camera mostly in Europe, definitely get a PAL version, and if you will use it mostly in the USA or to show your video to friends there, I recommend NTSC, because there are almost no TVs or monitors anywhere in the USA which can display PAL. Very few individuals in the US have a VCR which can record or play back PAL tapes. However, in most cities there are places which can convert tape standards, look in the yellow pages for "Video Duplication" or "Video Services".

Various places in the US, such as B&H Photo advertise PAL video equipment for sale. PAL equipment sold in the US almost certainly does not come with a US warranty, because as far as I know, Sony does not officially sell any PAL equipment here. You have to ask the vendor exactly what warranty you get, and what your service options are here, or if you live/travel in Europe. However, I don't think there are any "bad" Sony cameras, I think they are all made in a few factories in Japan and they are all of similar quality.
 

Focusing and Lenses

Yes, yes, and no. Yes, it has a manual focus ring on the lens barrel. In common with the VX1000 and XL1 semipro cams, it is not a true mechanical focus control, but is a continously-rotatable ring which is electronically coupled to the focus servomotor. The auto/manual focus slide switch actually has three positions. To infinity focus, you just pull down the switch past "manual" to "infinity", and release it (spring loaded, returns to "manual"). That sets the lens at infinity. By the way, if you use an external wide-angle or telephoto lens, the "infinity" focus setting may not be exactly at infinity any more (it all depends on the quality of the external lens). My 0.5x and 2.0x lenses do shift infinity slightly.

I've tried still pictures at every zoom setting, and even with an add-on x2 telephoto extender (Kenko VC-200Hi). The progressive scan stills look sharply focused at every setting (hence regular video, at lower resolution, will be as well). I have not observed the focus to drift as you zoom out from extreme telephoto.

With the camera zoomed all the way out (widest angle) you can focus up to about ½" (1.25 cm) from the front lens element. If you have the included sunshade/hood attached to the camera, this point is actually well within the hood assembly. With the camera at maximum telephoto, the near focus point is about 3 feet away. The remarkably close-focus feature mentioned above might be considered a bug when you are outdoors in bright sunlight and using the zoom at maximum wide-angle. The light causes the aperture to be stopped down as much as f/11 which can increase depth-of-field to include the front lens surface itself. Inevitably there is some dust there which the auto-focus circuit sometimes focuses on, instead of the real scene (a few people report this always happens; on my camera it can, but is rare). Solution: (1) when in bright sun, use the internal ND filter, optionally an external ND/polarizing filter as well (this opens the aperture, reducing depth of field; picture quality tends to be best around f/8 or f/5.6 anyway). (2) don't use full wide-angle, or (3) use manual focus. The zoom control is a rocker switch 1.25" long, larger than the one on my Canon ES2500. The harder you press, the faster the zoom. I don't have any trouble getting a slow zoom as long as I start to press slowly. You have to depress the rocker more than one millimeter before it starts zooming at all. The zoom speed range is spec'd to vary between 1.83 and 26.5 seconds. There are also aftermarket zoom controllers that work through the LANC jack, see my accessories page. Most people are more familiar with 35mm SLR cameras, so to review, here are a few standard 35mm camera lenses and their fields of view:
Focal Length    Field of View (degrees)  
20 mm            84  x 62   <- using 0.5x wide
28 mm            74  x 49
29 mm            64  x 45   <- using 0.7x wide
35 mm            59  x 39  
41 mm            47  x 33   <- TRV900 stock lens, wide
50 mm            40  x 27
105 mm           19  x 13  
200 mm           10  x 7  
400 mm           5.1 x 3.4
500 mm           4.1 x 2.7  <- TRV900 stock lens, tele
1000 mm          2.1 x 1.4
The TRV900 built-in lens has a 35-mm equivalent focal length of 41.3 mm (full wide) to 496 mm (zoomed in). (Note, the true focal length of the TRV900 lens is 9.6x smaller, as the CCD active area is that much smaller than a 35mm still frame.) You can see the resulting field of view from the above table. If you attach a 0.7 wide-angle lens, the focal length becomes shorter by 0.7, so fully-wide is 28.9 mm. If you use a 0.5x wide then the fully-wide setting is 20.6 mm. The formula generating the above table is: View_angle = 2 * arctan ( Film_dimension / (2 * Focal_length) ). It is not valid for lenses in "macro" mode, or for fisheye lenses. You can find much more detail about 35mm lenses (and nearly everything else) in the huge rit.edu Photo FAQ. I tried using the camera as-is, with no external lens or filter, by setting the camera at several distances from 6 inches to 2 feet from a ruler on my desk, zooming in as close as the focus would stay sharp, and noting the horizontal size of the full frame. Here's the results:

distance    h. size
------------------
 6"          1.9"
 9"          2.25"
 12"         2.75"
 24"         4.0"
-------------------

So, you can get a full-frame shot of a 4" long insect (are there any? preying mantis?) from 2 feet away, without adding any external lenses to the camera. If your insects are very shy, try a teleconvertor. With my Kenko VC-200Hi 2.0x teleconverter lens and using the camera at maximum telephoto, I also get a frame size of 4", but from 7.5 feet away! The resolution on this camera is very good, so you might be happy with the result even if the image was less than full-frame.

Meanwhile, I bought a set of 52 mm thread, +1,+2,+4 diopter "close-up filters" AKA macro lenses (more often used with still cameras). B&K Photo sells them, among others. The diopter rating is the inverse of the focal length in meters, for example, a 2 diopter lens has a focal length of 0.5 meters (which is 20 inches). The "+" means it is a converging (convex) lens. So, using a +2 lens means you'll need to be about 20 inches away from your subject (if you focus the lens at infinity, somewhat closer otherwise). You can also combine the close-up and telephoto lens for intermediate magnifications at a longer working distance. Carroll Lam (cassie at dakotacom.net) contributed the following:

Listed below is the nominal subject-to-front of lense distance and subsequent effective horizontal coverage for different combination of +diopter adapters. In all cases, the camcorder's lens is at max zoom. I was using the Hoya series of +diopter adapters acquired from The Filter Connection The measurements are given in inches (mm).
Lens combo Distance Size
+1 22.5 (572) 1-11/16 (43)
+2 12 (305) 1 (25)
+4 8 (203) 5/8 (16)
+4+1 6-3/4(171) 1/2 (13)
+4+2 5-1/2 (140) 7/16 (11)
+4+1+2 4-3/4 (121) 3/8 (10)

Here is a more comprehensive list from the same author, showing field size ("Image size") and focal length ("Subject-to-lens") at near and far focus points for every diopter-lens combination from +1 to +7 power, with all dimensions given in inches.

Diopter Lens Table


Usually the video cameras people mount on microscopes are "C-Mount" type which have removable lenses, so you rely on the microscope column optics to actually focus the image on the CCD surface. The TRV900 has a non-removable lens. What this means is you will need collimated light out of the microscope, instead of focused light. This is the same optical property as an eyepiece, since your eye also takes collimated light in and focuses it internally on your retina. I think that collimated light out of a microscope camera port may be unusual, but if you could use the normal eyepiece, and you can get or make an adaptor to go to the TRV900's 52mm thread filter mount, you'll probably be ok. There may also be a concern about the weight of the camera supported only by the lens mount (it's not designed for that) so there may be some mechanical support difficulties to overcome. I've seen a third-party microscope/telescope adaptor for video cameras which might help you, look here.

Note also, this camera works best with outdoor daytime light levels. I thought microscopes often work at very high f-numbers in the optical system meaning you may need very high illumination levels for the best image, but I'm not very familiar with the details. If you can use a "standard" color video camera on the microscope, the TRV900 ought to work as well, light-wise. Well, if you have slow-moving subjects you can use a slow shutter speed too (down to 1/4 second), for better images in lower light.

Measured TRV900 focal length Jake Socha (jjsocha midway uchicago edu) Sept. 25, 2001
When I assume that the widest angle is at 4.3 mm, the estimate for the max zoom is 48 mm, instead of the [Sony] listed 51.6. When I use the zoom as the reference, the wide point is estimated at 4.6 mm. Both answers are about 7 % off. I don't know which to use [exact CCD chip dimensions are unknown], but I choose the regression with the wide as a reference, because most of my unknown focal lengths are closer to that. Here is a graph of focal length vs. zoom setting. The graph of 'clicks' vs focal length is only approximate. The zoom button on the RM-95 seems to be extremely sensitive, and no matter how quickly and evenly I pressed it, I got wildly different numbers of clicks between wide and tele, ranging from 70 to 104. The shape of the graph, though, I thought to be interesting. Here's my formula for focal length in mm:
f = 0.00494126 (mm/px) * distance to object (m) * Object size in image (px) / Object size, reality (m)
-Jake

Outdoors and Underwater

The camera is only spec'd to operate down to 32 F (0 C) but I've found it is better than that. I was in Ohio in late December 1998 and I took a half-hour walk outside with the camera at 14 deg. F (-10 C) and shot some landscapes, rocks, deer etc. There were no problems, except that with heavy gloves, it is difficult to press the PHOTO button without also hitting the zoom rocker. You'll probably see somewhat shorter battery life than at room temperature. I put the camera in my camera bag before re-entering the house, to let it warm up slowly and avoid water condensation. After half an hour I took it out and everything worked fine.

For colder weather you may want to investigate chemical heat packs, like skiers use to keep their hands warm. The one commercial heated jacket I know of, the "Polar Bear" from Portabrace, is only for large shoulder-mounted cameras. Howard McCollister posted the following to rec.video:

I live in central Minnesota, spend a lot of time outdoors in the winter snowmobiling/skiing. Minus 30 C is nothing new to us here in the land of Jesse (The Mind) Ventura. I have had good luck in the cold with electronic devices such as GPS, digital cameras, or camcorders by attaching 2 or 3 disposable heat packs with rubber bands or tape. They'll last several hours and work well. Don't forget a few for your boots. Tape one to the battery, and one on either side of the camera. You have to be careful when you bring the thing inside, however. You WILL get a "dew" indication. Take the tape out before you go inside. These packs are cheap, and completely biodegradable.

I have a friend that puts his Handycam inside an electric sock, the kind that work off a D-cell battery. He cuts a hole for the viewfinder, and a slot to slip his hand onto the grip. Seems to work very well. A lot cheaper than the commerical camera muffs...

H. McCollister

The TRV900's CCD chips are very sensitive to light. This is good for indoor settings, but outdoors when it is sunny, you need to use filters to cut down the light for best results. Otherwise the recorded image has a washed-out appearance lacking in contrast and color saturation. The camera has an internal ND filter (Neutral-Density: a neutral gray, not colored) which is activated by a button near the lens, but this is usually not adequate on a sunny day. You ALSO need to use an external ND filter which you thread on the front of the lens. You could substitute a polarizing filter, if desired; see my filters page.

Note that I'm talking about the recorded video image; regardless of your lens filter, of course the fold-out LCD display is also washed out by sunlight, which you can improve by using a shade such as the "Hoodman" product.

There are probably two problems happening at once here: (A) too much light saturating the CCDs, and (B) atypical subject contrast, fooling the autoexposure circuit. As for the first item, several people have emailed me to say the camera has a washed out picture in bright outdoor settings- I'm sure the sun on snow, or desert sand, is about as bright as it gets. The solutions, choose any or all, are as follows:
(1) Use an external filter. I use a polarizer (-1.5 stops), but others use external ND filters ("Neutral Density" filters appear grey or silvery, and cut down the incoming light level without altering the color balance).

(2) Turn on the camera's internal ND filter (about -2.5 stops).

(3) Set the internal gain down (menu selection GAIN SHIFT: -3db).

(4) You can manually increase shutter speed, and you can enable the menu option (AUTO SHTR: ON) that allows the camera to automatically increase shutter speed in bright conditions as necessary, but this has the usually annoying side effect of giving a "strobe-effect" to fast motion. However, fast shutters are useful if you want to extract non-blurry still frames later.

To address issue (B), atypical subject brightness, you should know that autoexposure circuits on video and still cameras are set for "typical" scenes. This turns out to mean an average reflectance of 18% of incident light. Your snow-covered hills are much brighter than that, but the camera doesn't know that, so it tries to make the snow look grey, and that makes anything darker than snow, look black. The remedy is to use manual exposure- turn down the aperture until the subject of interest looks properly exposed. One "pro" approach would be to use a 18% grey card. First, set the exposure control switch (labelled "Auto Lock/Hold") to the middle position (auto lock release) but make sure no exposure adjustment is active (no indicators on in the viewfinder). Now fill the camera's field of view with the grey card (illuminated as the subject will be). Now, press the "exposure" button (but don't turn the adjustment wheel!). You'll see the camera's automatic aperture setting (f/stop) appear in the viewfinder. Now pull the auto lock switch down to the "Hold" setting and you'll have the correct exposure locked in. Repeat the above if your lighting angles change, clouds move in, etc.

Some pro video shooters prefer to just set the manual exposure by eye, based on contrast judgements in the viewfinder, but I feel this takes real experience- playback on my TV usually looks somewhat different than it does in the viewfinder. You can use the "zebra bars" menu option to tell you which areas will be overexposed (pure white) but again, it's a judgement call how much solid white you want in the frame- for a dim interior, probably none, but for a snowy scene, it might be quite a bit.

Yes, there are many underwater housings available from many sources, check here.
 

Distributing/Archiving Movies

I'm not a professional or even very advanced amateur, but... I think the camera is more than adequate for VHS distribution. With well-lit scenes the clarity is breathtaking, and the colors are excellent- far beyond what you would ever see in a VHS dub.

I do not have experience with broadcast production*. It's hard for me to imagine NTSC video looking any better than this, but I haven't seen what a pro camera on a studio monitor looks like. All I can say is, if you've got good light, I would be very surprised if anyone objected to the image quality.

*Well, I was an extra in a CBS (?) "Movie of the Week" about eight years ago, if that counts. They shot on film (35 mm I think) and the amount of money they lavished on that production was stunning to behold, and a bit sad, since the movie was entirely forgettable. Part of the camera is visible behind me in this shot of some of the extras.

One film student told me he printed some TRV900 footage onto 35mm and it came out well (See "62) Comparison of video and film" here). However, I think the (non-professional) video camera which has been used most for this kind of project is the Sony VX1000. You can turn off the edge enhancement on it, which is apparently important for the best looking film transfer.

If we are talking about a feature-length film you should realize this costs several tens of thousands of dollars. Transfer of a 90 minute feature to 35mm film runs about $35k from DVFilm or Takafilm, or $52k from the Sony High-Definition Center. Victor Khong lists some more companies on here. If you are talking about this kind of money, I think an extra $2k for a better video camera seems reasonable.

[1/28/01] The DV video format which the TRV900 and all other MiniDV and Digital8 cameras use, runs at a fixed data rate of 3.6 MBytes/sec. That is 216 MB/minute or 13 GB/hour. This rate is much too high for video you intend to put on the net or on a CD-ROM, so you need some more compressed format. Note, while it can be done, I do not advise emailing video at all; better to put it on a website somewhere and just email a link to it. If you must email, check beforehand to see if the recipient can receive large files in email. Some people, myself included, still use 56k modems and don't want to wait 20 minutes just to download their mail. Some ISPs restrict email to 1 MB files, or less.

There are many different video compression formats available to use for your web or CDROM production. For CD-ROM applications, I would recommend MPEG1, one of the oldest compression formats. Quality can be very good, and MPEG1 viewers are already present on most machines. There may be advantages to MPEG2 but decoders are less common. The new MPEG4 format may give the best quality per megabyte, although I've had trouble getting this format to play properly on a slow machine (640x480 playback may need more than a 500 MHz P3 anyway). For web pages, I'd consider MPEG1 for downloads, or the real-time streaming formats like ASF (MPEG4), Real Video, and Quicktime. By the way, MPEG-4 is a complex format that encompasses much more than just video, see the MPEG-4 Standard.

The original DV is 720x480 pixels, 30 fps (NTSC) or 720x576 pixels, 25 fps (PAL) but for these applications you will end up using lower resolution and often lower framerates to match the abilities of CDROM drives and the Internet to deliver data, and the host computer to decode it. "Streaming" web video shows up in real-time on your web browser. It is very highly compressed and this has consequences for quality. As you'd expect, streams at ISDN or T1 datarates will look better than those intended for 56k modems. If you want to see some examples online, check out the movies at ifilm.com.

Ulead MediaStudioPro (Ulead MSP) 6.0 can directly create MPEG1, MPEG2, Microsoft ASF and RealVideo formats from your DV format file. I believe Adobe Premiere 6 can also. Some MPEG1 and MPEG2 creation tools are free downloads, for example TMPGenc and bbMPEG. If you have a Canopus DVRaptor or other Canopus product, you can use their software to convert to the ASF format.

[3/21/01] Microsoft makes a free MPEG4 encoder. Microsoft's version of MPEG4 is called ASF (advanced streaming format). As of March 2001, the Microsoft ASF encoder is called Windows Media Encoder 7. It generates files which can be played back by the Windows Media Player 7. Both of these are free downloads from this page.

There are various other MPEG4 encoders, including several variants called "DivX", which may or may not be compatible with the MS Player- it does not work with version 7 of the player, I think. I don't know if these other encoders have any advantages. DivX seems to be most often used by those making copies of DVD movies. There is some DivX information at this page and no doubt your favorite web search engine can find much more.

There is a lot more to the topic than what's mentioned here: if you are investigating video formats to produce your own multimedia CDROM or website, I have found Terran has a very useful Codec Index at their Codec Central web page, which lets you know what's available in this field. There is a magazine called Streaming Media devoted to the topic.

Yes. The easiest way is with a Terapin VCD recorder, which works like a VCR, and can record VCDs in real time. Or, if you have a CD-Rom burner, either an analog video capture card or firewire card, and some software, you can make one on your computer. Remember that VideoCD uses MPEG1 at 352x240 pixels NTSC (352x288 PAL) and is often called "VHS quality". It is not to be confused with DVD. I have made some VideoCDs and found the image quality for stills is comparable or a bit better than VHS, but there are blocky artifacts on almost all scenes with motion that are quite obvious on anything but the smallest TV (however played at 1:1 on a computer monitor, they look reasonably good.)

A VideoCD won't give you DV quality, but there's little question that CDs (even writable CDs) are more durable and long-lasting than VHS tape (whether the appropriate VideoCD playback machines will be handy in the future, is another question). Be aware that to play your VideoCD, you need a player which (1) reads VideoCD (some DVD players do, others don't) and (2) can read the CD-R or CD-RW media you wrote it on (many DVD payers don't). I have owned a Pioneer DV-414 which played VCD on CD-R with some dropouts, and a Philips DVD 825 and Afreey LD2060 which could play VCD/CD-R with no problems. If you get into VideoCD burning you may find it a maze full of twisty passages, all different. To aid you, there are MPEG1/VCD FAQS, also a VideoCD making guide and another VideoCD FAQ. I have found the comprehensive CD-R FAQ and www.vcdhelp.com very useful as well.

By the way, Adaptec EZ CD Creator Deluxe gives you a basic VCD making tool but if there's a full-featured VCD authoring package aimed at the consumer, I don't know about it. VideoPack 4.0 from CeQuadrat retails for US$990.00. "MPEG Power Professional" (US$500) is supposed to be a good MPEG1 encoder from heuris.com. You can download bbMPEG and AVI2MPG2: these are freeware Windows programs that convert AVI files to MPEG-2 or MPEG-1 (including VideoCD and SuperVCD). Some claim that de-interlacing first with Virtual Dub improves VCD quality. I find TMPGEnc works very well for converting DV to MPEG1 at VCD or higher bitrates. If you want to try newer and less-standard video codecs, Media Cleaner EZ4.0 is a free download (for Windows).

Subject: Re: Jump to DVD (how about SVCD?)
Date: Tue, 11 Apr 2000 16:12:20 -0400
From: "Harry S"
One cheaper solution than the DVD-RW solution mentioned is SVCD. SVCD is a format orginated in China and is basically MPEG-2 video at 480 x 480 NTSC at bit rates up to 2.5Mb authored onto a regular CD or (or very cheap) CD-R. This allows for about 40 min. of video on a CD at the higher data rate. (The 480 x 480 is stretched automatically in SVCD players to standard 4:3 aspect ratio NTSC)

I have been experimenting with this format, and it seems to be a pretty good compromise. It is far better than standard VCD format, but not quite as good as DVD. I would rate it about a very good SVHS quality.

What I have done is use DV footage from the TRV900 captured in Premiere and using the freeware program bbMPEG (available as a Premiere plug-in) produce MPEG-2 video and audio streams compliant with the SVCD standard. Then I use the program I-Author by EnReach which produces an SVCD image file that can burned on to a CD-R using Adaptec CD Creator. (Unfortunately, no Mac solution yet to author SVCDs)

How do you play SVCDs? Well there's a catch here! Most DVD players won't, but a super cheap DVD player from China introduced in the U.S. does - the Apex AD-600A (at about $170). There are some programs which allow you to play SVCDs on PCs as well. There are also stand alone SVCD/VCD players available in Europe and Asia too.

Just a far cheaper alternative I would suggest looking into. More information on SVCD from philips.com (.pdf documents from the Philips web site)  - Harry S.

As of April '00 the basic I-Author package costs US$1000. If you use Linux, you might want to try VCD Tools which reportedly can create double-bitrate and SVCD as well as normal VCD formats, and is free. That webpage claims "most" DVD players can play double-bitrate (2500 kbit/s) VCDs which have better video quality than the standard 1152 kbit/s MPEG-1 file in a normalVCD. I like TMPGEnc (Tsunami MPEG Encoder) for MPEG1 and MPEG2 encoding. I found this TMPGEnc/SVCD page a useful additional resource. Nero 5.0 from Ahead Software will make a SVCD format disk from a compliant MPEG2 file. Jukka Aho wrote a useful SVCD Overview including links to other SVCD authoring software. There is a useful page regarding TMPGEnc for SVCD making here. You can read some of the standard specs in this Philips SVCD Technical Note. Not all DVD players can read SVCD; I think most older models can't. The Raite AVPhile 715, the Apex 600A, the Afreey LD2060 and Sampo DVE-620 (see below DVD notes) are a few that can. Here is a AV715 user page. There is a large VCD/SVCD/DVD-R/DVD-RW compatibility list at www.vcdhelp.com. People are saying good things about the Dazzle DVC II, a card that can record video in MPEG1 or MPEG2 format for burning to VCD, SVCD or (in theory) DVD format disks. 'Heyday' has a user site for the card. The Amoisonic VDR2000 is a hardware box which can record VCD or SVCD in real time.

[4/3/02] Yes, and DVD is the only distribution medium that can display the full image quality the TRV900 is capable of. (MiniDV isn't good for distribution- if your client has a MiniDV camera you can try sending a tape, but even in SP mode, some dropouts may occur when playing back on a different machine. Digital8 in SP mode using good Hi8 tape stock may be better.) You can now buy a DVD-R writer for around $400. The first one on the market was the Pioneer DVR-A03 DVD-R and DVD-RW writer (their current model is the Pioneer DVR-A04). I believe the Pioneer drives are used in all Mac computers with DVD writers, and they are sold separately for Windows PCs as well. I have one of the DVR-A03 drives, and I write about DVD authoring with it here.

There are also stand-alone DVD writers which do not need a computer, in theory these are easy to use like a VCR. One such product is the Pioneer DVR-2000, see the description here. You can buy DVD+R hardware also, but the advertising hype claiming these disks were "more compatible with existing players" than DVD-R doesn't seem to have panned out.

There are several companies offering to create a DVD for from your MiniDV tape (or SVHS, etc.), for example Jason Foodman at SwiftDVD, Ian McDermott at BT Media, and Excel Productions.

[12/29/00] There are some inexpensive first-generation DVD-RAM writable drives out there (2.6 GB per side) but they are for data. They cannot create a standard-format Video DVD (4.7 GB single-sided) which a set-top video DVD player can read. If that is acceptable, maybe the Canopus Amber would be of interest. It is an integrated system for producing MPEG2 video on DVD-RAM, including real-time MPEG2 encoder hardware.

Mini-DVD If you don't have a DVD-R drive, there are a few DVD players which can read the DVD filesystem with MPEG2 video content off a regular, inexpensive CD-R or CD-RW disk, meaning you can make your own "mini-DVD" without an expensive DVD-R writer. However, this is the realm of a few brave experimenters with time on their hands right now; no company is supporting this format. Note; at 9 Mbps, you only get about 10 minutes of DVD quality video on a regular CD, and typically only a specially "hacked" DVD player can play it; see below.

The Afreey LD2060 can play CD-R type MiniDVD discs, and is around $250. The Sampo DVE-620 is basically the same, and there is a Sampo user's forum for it thanks to nerd-out.com. There is more about MiniDVD making at Robshot.com and Bjorn's page. In stock form, I believe these players reliably play bitrates from CD-R only to 2 Mbps which is rather low quality: the standard DVD rate is 9 Mbps. Some intrepid experimentors have replaced the standard EIDE DVD-ROM unit in the Afreey or Sampo DVD player with a faster one, eg. the Afreey DD4010E.

I have the LD2060 and it plays a "miniDVD" I made with TMPGEnc (MPEG2 encoding), DVDMotion CE (DVD authoring), and Nero 5.0 (CD-R burning). My MiniDVD disks play smoothly only at 2.5 Mbps rate; higher bitrate files skip and freeze-frame, and when I tried to upgrade the LD2060 with two different faster EIDE DVD units, it didn't work at all.

The DVD format employs MPEG2 compression (4:2:2 sampling, interframe motion estimation etc.) which is more complex than the DV format used by the TRV900 and NLE systems based on DV. Commercial MPEG2 encoding is done in hardware, for speed, but there are software MPEG2 encoders, for instance from Ligos. You can find out more about DVD hardware from brouhaha.com or visit the DVD FAQ. This DVD Page is full of useful info as well. DVDIt.com sells DVD authoring software and has some related information. SpruceUp is a fairly cheap DVD authoring program.

DVHS or "Data VHS" can record MPEG-2 data in digital form in several different formats including high-definition or HDTV formats. Examples of digital video programs are digital cable and digital satellite TV. As far as I know, it is not compatible with any currently existing DV cameras in any way. You can read more about the format at DVHS.co.uk. Panasonic sold a number of PV-HD1000 DVHS decks, it was the first on the market, but apparently is now claiming that model does not exist (see this page.) If you want to record DV, Panasonic sells the AG-DV2000 which is a DV/MiniDV VCR, which is compatible with DV cameras, now selling (Jan.2000) for slightly more than a TRV900. For more on this and other MiniDV decks, see this listing. The TRV900 works well with the Mac G3 with built-in firewire, according to reports, and the Sony VAIO laptops (built-in firewire) as well. The very first Mac G3s in Feb. 99 had an incompatible firewire driver but this was fixed in March 1999. The TRV900 has been out for over a year now, and is a popular camera, so any computer sold for video editing ought to be compatible. If that doesn't describe your system, you will have to research the hardware you have or need to get. For further details and links proceed to my DV Video editing page. I use the Canopus DVRaptor card which I like. Previously I used the Bravado DV2000 which I didn't like as much. There are a lot of firewire interface cards on the market, most of which are PCI cards to plug into a PC. There are a few PCMCIA (PC-Card) format devices for notebook computers as well. The best place to find information on these cards is Pat Leong's NLE page, and there are some comments on my video editing page. Look at the rec.video.desktop newsgroup also for ideas. Post your question there and I'm sure you'll get some opinions.

Technical Details

Depends how you measure it, evidently. I will pass on several different answers: this page reports data from the Swedish magazine Ljud & Bild. I have below a note from Giles Read in the UK, a newsgroup report of a published test, and some images from a waveform monitor showing a 5 MHz sweep. I have my own resolution test and I compare the TRV900 recorded image via firewire, via analog in, and to/from SVHS and VHS formats here. I measured 540 lines of resolution for firewire transfer, and 480 for analog in+out.

Note 0: European cameras use PAL format which generally means higher resolution than the US NTSC format, but for digital, I think both of them have 720 pixels across (?). Note 1: do check out the full review in Sound and Vision; it is extremely positive- they seem to like the camera even more than me :-). Note 2: if you would like a refresher on how video resolution is defined, look at this EliteVideo site.

(...) Shortly after I got the camera I spent some time with a friend who is a professional video engineer, and we compared the '900 to a professional studio camera that he just happened to have handy. The '900 acquitted itself quite well! We estimate that the studio camera had a resolution of some 700 lines, while the '900 came in at around 550 for full resolution, with a limiting resolution of about 750 lines. The VTR part of the deal was still giving us resolution at 5.8MHz - the limit of the measuring equipment to hand.

-Giles Read

Lab Test Gives TRV-900 High Marks
From: bigrocketman3@webtv.net
Newsgroups: rec.video
Subject: Sony TRV-900 Test Report
Date: Tue, 19 Jan 1999 14:40:20 GMT
Confirming the good performance that many people have described on this group, Sound & Vision Magazine gave Sony's 3-CCD mini-DV TRV-900 camcorder a top rating. In its premiere issue, for February-March, Cliff Roth gives one of the most extensive and useful evaluations I've seen for a video product. Lance Braithwaite, who has provided Video magazine with good, professonal evaluations for many years, conducted the laboratory tests.

The tested level of resolution from the TRV-900's camera section was 530 lines, with 500 lines viewable in tape playback. The color viewscreen displayed 410 lines of resolution and the small viewfinder gave 380 lines. The signal-to-noise ratios were the highest I've seen from anything less expensive than a broadcast camera. The weighted luminance reading, from the S-Video output, was -60 dB and the chrominance PM was -56.6 dB. These figures came from the playback of footage recorded from the TRV-900's camera. When a pre-recorded reference tape was used, the readings were -66 dB and -63.6 dB, respectively.

(...) The entire TRV-900 test report and other articles can be seen online at Sound & Vision's website: www.soundandvisionmag.com.

Steve McDonald

The PAL and NTSC versions of the TRV900(e) use different CCD chips, labelled ICX216AL-13 (NTSC) and ICX217AL-13 (PAL). The pixel count is listed Sony's web site as: NTSC: 380k pixels (340k effective), PAL: 450k pixels (420k effective). This is the pixel count per chip: each camera has three of these chips, corresponding to red, green and blue color sensitivity. The 3-CCD design gives you much better image quality than a single CCD camera would have with as many pixels.

You might guess that Sony wouldn't add more resolution to the camera than the DV format can record, but it seems that they have. For the NTSC case, each CCD chip has 34400 extra pixels beyond what's needed to match each of the 720x480 DV format pixels with a CCD pixel. This is about 71 extra pixels per scanline beyond what the DV format can actually record. For the PAL case we have 35280 extra pixels above 720x576, or about 61 extra pixels per scanline. So, why is that?

Normally, CCD chips contain several extra rows of non-imaging pixels which are physically masked off (eg. with an aluminum layer) for dark level subtraction, to correct for voltage offsets due to process variations at manufacture and temperature variations at time of use. Here, I think there are more extra pixels than you'd need for that purpose, which is why I surmise that there isn't a 1:1 match between the CCD and the DV pixel map. Various tests from myself and others indicate slightly higher resolution from the camera's outputs "live" vs. in tape playback, which seems to agree with this interpretation. I think the CCD output (in discrete pixels) is converted to analog video (continuous scanlines), and then internally re-sampled into 720 pixel scanlines for DV recording and firewire output. In theory, the more pixels in the original signal (from the CCD), the better contrast you get on thin lines and detail, and the less obvious jaggies/aliasing you see when panning or zooming slowly over such detail, even after downsampling to a fixed 720 pixels per line.

According to my measurements, the camera can distinguish detail over an intensity range of about 50:1 which is almost 6 stops. My measurements are described here. I understand that professional video cameras are about 1 stop better, and the cutting-edge HDTV cameras can hold 11 stops, which rivals film in exposure latitude. Recording with the internal mic and listening by ear, I don't notice any problems; but if you like numbers you won't like these. First the background: I used a Turtle Beach Fiji soundcard and SpectraLAB software to generate a 1 kHz sinewave for recording, and then analyse the playback using a fourier transform. The test setup itself has stereo 16 bit D/A and 20 bit A/D, and has 0.008% THD+N and 82.3 dB SNR in analog loopback mode. In the table below I show what I measured for all the audio recording devices I had handy today. Note that what I am calling SNR is actually the signal to (broadband noise+60 Hz pickup+distortion) ratio, and it is a more stringent test of an audio system than a simple dynamic range measurement.

With 16 bit sampling at 48 kHz, the MiniDV format should be capable of reproducing CD or DAT quality sound. I tried two different signal generators, with and without attenuation right at the mic input. The TRV900 consistently comes in at 0.8% THD+N and 42 dB SNR, which is surprisingly low. It doesn't matter if you use 12 or 16 bit mode. I think the culprit is the microphone amplifier (line in is available only for 2nd track dubbing, or external source video recording). I got the same numbers if I use headphone or A/V as audio out, and at several different record level/input level settings. 60 Hz + harmonics is the dominant noise term, but not by a large amount. Sony doesn't spec audio SNR or THD on this camera. If you want audiophile sound, better use a separate minidisc or DAT recorder. Note, by the way, "dynamic range" is quite different from SNR and is probably a much higher number, but I didn't measure it.

The A/V line input of the TRV900 tests at 43.4 dB SNR. However, the dominant "noise" is 60 Hz and 180 Hz (3rd harmonic), which suggests hum pickup in my cable. If you ignored those two frequencies, the figure would be about 63 dB SNR. I actually tried testing the internal microphone, recording a 1 kHz tone from a speaker. That gave me a reading of 35 dB SNR, but part of the noise was a 15.8 kHz spike, which was no doubt the horizontal retrace from a TV which I had on and nearby.

It's difficult to test these things properly. My SNR measurements are considerably lower than the manufacturer's claims, in every case; perhaps I'm doing something wrong. All I'll claim that you can really infer from this chart is the relative rankings of audio performance, since I used the same test setup to measure each one. It so happens the ranking by SNR agrees with my prejudices about sound quality, in order: Turtle Beach Fiji soundcard, Sharp 702 minidisc recorder, Sony TRV-900 DV camera, Panasonic AG-1980 SVHS deck, Sony WM-D3 Dolby-B recording walkman, and last and least, an inexpensive dictation cassette recorder.

                    THD %   THD+N %  SNR dB    signal dB


TB Fiji   line in:  .002   .005      86.1     -3.7
   "       mic in:  .007   .053      65.5     -15.4
Sharp 702 line in:  .027   .031      70.3     -4.5
   "       mic in:  .037   .057      64.8     -5.8
TRV-900   line in:  .030   .673      43.4     -9.7
           mic in:  .345   .800      41.9     -23.1
AG-1980 SVHS deck
HiFi      line in:  .036   1.24      38.1     -4.2
linear    line in:  1.24   3.44      29.2     -4.5
WM-D3 Pro walkman
Dolby B   line in:  1.36   5.15      25.7     -6.3
Panasonic RQ-L340
dictation recorder
(mono)     mic in:  0.81   11.1      19.1     -5.1

Remember to take these numbers with a grain of salt. Subjectively the camera comes off very well: several people have emailed saying the camera has great sound and they've made their best-sounding audio recordings to date with it. One Usenet poster did say he found the Sony VX1000 to have better sound, but another user wrote that the TRV900 sound beat out the VX1000 (see the comparisons here.)

[7/24/2004] Initially, there was no known solution. More recently the answer is yes, the TRV900's audio performance can be significantly improved by a modification that bypasses its internal mic preamp. Damir Vrancic modified his camera as described here. He just bypassed the first two gain stages (BA7780 and M5222 chips) after the external mic input, with the effect that gain is fixed (roughly mid-range) but he got 14 dB better nois