DV v. D8 (again)

[A_BIT]

Show us a mini-dv cam that can deliver 500 lines through the lens to tape within the low-end D8 price range.

[Brian Ellis]

DV v. D8 (again)

Just to add fuel to the fire, I quote from p. 56 of the November issue of Camcorder and ComputerVideo:

"It appears that luminance resolution also is better in pure DV cams than in present D8 or Hi8 units, probably in part because of the higher pixel count in DV CCDs. The current Sony D8 cams seem to draw heavily from Hi8 units also in the 1999 line, Apparently lenses and CCD pixel counts are similar and most probably many of the mechanisms and electronic chips are shared across the product line.

"It might be interesting to reflect on what might happen in D8 gear if their CCDs were upgraded. Present D8 units seem to fall a bit short of reaching the 500 line potential of DV when taping using the cam's own lens and CCD. AGAIN It is interesting to find Sony stating that 'maximum resolution is achieved only when recording through the video inputs,' This would seem to indicate that the D8 VCRs can attain true DV specs and Sony's idea of recording at 4500 head rpm on Hi8 tapes has considerable merit."

I'm only quoting, so don't jump down MY throat, please

I recently spent a few days in Seattle so, as a postscript, I'd like to make a few remarks about what I saw there. Camcorders are generally about 25-35% cheaper than in Europe for comparable models, even a PAL model I saw in one place. The numbering of models seem similar for Sony, so comparisons were easy. Canon seem to prefer names to numbers but it is Panasonic who really confuse the issue in that the model range seems entirely different for most of the models and, even where they are similar, the numbers aren't. I was REALLY astonished to see how little DV/D8 has penetrated the US market, judging by a number of shops' offerings. In the Far East, it is difficult to find analogue camcorders and this is also beginning to be the case in Europe. In one shop in Seattle, I saw just one Sony DV and a Panasonic and a couple of dozen analogue models. I also went into a CompUSA store and noted they had a special counter labelled "Digital Video" but this was mostly devoted to still cameras. Again they had two DVs, no D8s and 6 analogue models on display. They had no video cards nor NLE packages. Really hard to justify the label. In fact, I was really disappointed with the shop as a whole for several other reasons. Let me recount one off-topic incident. One of the reasons I went there was to buy a FrontPage 2000 upgrade. I found the box with a sticker on it saying this box is empty, please consult the sales staff to obtain the software, which I did. I was told that they do not stock that item. When I remonstrated that they should withdraw the boxes from their display shelves, I was told that they could not, because Microsoft paid them to have the boxes on display. I pointed out that this was tantamount to fraud as they were obtaining money for a service they could not fulfil. I wasn't quite thrown out.

Anyway, CompUSA is now in my very dark grey book. Incidentally, I found my upgrade in a neighbouring Barnes and Noble store, $2 cheaper than the price marked on the shelves at C-USA, so the latter did me a service by not stocking the item.

------------------
Brian (the terrible)



[This message has been edited by Brian Ellis (edited 06 November 1999).]

[VIKINGMAN]

Well Brian, I have to agree what their test showed and what they've stated about the D8-format, also on a longer temr basis, the Sony-drives are not build to cope with the high speed of the Digital8 tape- as mentioned in the other thread - a patched solution from Sony, to keep their costs (and prices) down, but NOT to follow a stadard made by 80% of all other electronic manufactures!

------------------
ASUS P2B-S, PII-350 (o/c to 412MHz), 128MB RAM, Cheetah 9.1 GB, Matrox Mill. G200SG, SB 64AWE, Plextor 32x CD-Rom, Sony CDU-924S CD-R, Canon BJC-7000 InkJet and Canon CanoScan 300 Scanner.

[Brian Ellis]

I agree with A_BIT and partially with Grigory. As I said, I was only quoting.

As for Grigory's statement about 3 CCD cameras, he should know that some (at least Panasonic, who developed the technique, and Sony on latest cameras but not on earlier VX series) have an astuce to provide a higher luminance resolution. The signal is derived from all three CCDs which are mechanically staggered slightly, so that the luminance signal of each is 1/3 pixel different from each of the others. Of course, the horizontal resolution does not approach the 1300 equivalent-lines that this method could theoretically offer because the bandwidth must be restricted to avoid overlapping the chrominance signal on the combined result. I'm not sure what resolution is actually obtained but, in any case, this is a sales argument because the lenses are inadequate for much more than 500 lines, even on the Sony VX-9000. Perhaps the Canon XL-1 is better in this respect???

Furthermore, the lens quality does play an important role, even on low resolution cameras. I remember Ilford used to demonstrate this very effectively, using three monochrome films in 35 mm still cameras, Pan F, FP-3 and HP-3. The Pan-F was an ultra-fine grain film which they developed in their stadard ID-11 developer, which was used for all the others, as well. This developer was a good all-round fine grain but not high-acutance developer. They shot test patterns on all three films with an Asahi Pentax SLR using an f 1,8 Takumar lens and a lower quality unnamed lens at various apertures. If my memory is right, the optimum resolution for both lenses was at about f8. With the Pan-F, they measured the resolution (I forget the lines/mm but I think that, at the centre, it was about 400 and 200 for the two lenses). The FP-3 film was able to resolve about the resolution of the poorer lens and the HP-3 considerably less. The results showed that, with the FP-3, the lens quality still played an important role and, even with the grainy HP-3 (it was the fastest film on the market at that time, in the early 1950s, much faster than Kodak's Super-XX), which was nowhere near the resolution of even the poor lens, the good lens gave a better image. I cite these tests to demonstrate that the bottle-ends found in cheap cameras will certainly degrade the image beyond what a good lens would resolve. A little thought will give the reason (compare with the variable dot sizes in laser and ink jet printers which give an apparent sharper image at a given dpi).

As for the 1-CCD models with the larger CCDs, I can quote Panasonic NV-DA1 and NV-DS77 and the Sharp VL-PD3.

I'm pretty sure that it is technically feasible today for a semi-pro 3 CCD camera to come up with the ideal horizontal resolution (PAL) of 800 equivalent-lines, with an excellent lens and the rest to match, but the electronics would be more complex. The cost: anybody's guess but at least 2-3 times an XL-1 or a VX-9000 -- and then what would we record it onto, a new super-DV format???? It would require twice the bandwidth of current cameras.

Anyway, it is interesting that two magazines in the USA and the UK have independently come to the same conclusions about this matter of DV and D8, despite what your theory says, Grigory.


------------------
Brian (the terrible)

[Grigory]

Brian,

I had no theories, just numbers from Sony.
Yes, I have no idea how to test the lense quality of D8.
What I can, is to read the pixel numbers.

As for "independent" magazines, their references to "lower numbers of pixels" for same 1 CCD devices look very "independent" from official Sony specifications

It seem to me that lense quality just fits the CCD requirements, both for mini DV and D8. At least for Sony, it would be strange to use different (by quality) parts such as lenses. The cost of camcorder lense, in case of consumer camcorders, should not be a problem. Any photo camera can easily overcome camcorder in terms of resolution.
As for 500 lines, it is theoretical limit of DV format itself. Near this value, even the highest resolution CCD and lenses will produce moire. The picture quality may become even better with lowering resolution.

I agree that the lense resolution is only one parameter from many. The others, such as all these distortions and aberrations are very important too. I expect that here you can find a lot of differences.

Note, Sony has a small asterisk near every camcorder at the resolution row in a table.
I am asking: why "two independent magazines" took the asterisk only for D8, but not for the other camcorders? The footnote says that 500 lines are achieved for DV format video, transferred by firewire.

If another manufacturers did not place similar footnotes in camcorder specifications, does it mean they have 500 lines of resolution for 500 pixels per full line on thier CCD? This is impossible. So, the absence of footnote is simply a marketing move.

Grigory

[This_Idiot]

Nice analogy Brian.

Did things move on ? I used Pan-F, FP4 and HP5 when I last used my own darkroom - Pan-F was absolutely awesome, even on my old Halina viewfinder camera.

Interestingly, I had 2 enlarger lenses which I used interchangeably. The Nikon was as sharp as a razor but very unforgiving on blemishes (funny, doesn't matter how much wetting agent you use there's always some drying marks). The Neotaron was well named, it was trying to shine a light through tar. But for developing portraits it was ideal, built-in softenening and disguised a host of horrors.

[Grigory]

You make me laughing!

Ten years ago Communist Party of the Soviet Union said: "It appears that the life in the Soviet union is better than in USA...."

To check this, we had to travel to USA, which was not allowed for most of us...

Now, ALL you can "travel" to Sony site and see that the NUMBERS of pixels are EQUAL for 1 CCD mini DV and D8 camcorders: 800,000. So, the resolution is NOT 500 lines for all consumer devices from the CCD part!

There is only one exception, appeared AFTER introduction of D8 : a new megapixel CCD mini DV camcorder, at much higher price, of course. Even 3 CCD units have less than 500 lines luma resolution, but better colors, of course.
So, all "resolution and quality" comments are simply not true. The phrase about 500 lines is a ritual phrase used in every manual of DV camcorders.

800,000 pixel CCD with 400,000 active pixels can provide 420 lines of resolution from lense. I measured this by myself, as well as many others did this too. This value fits the theoretical limit of such CCD. So, the lense has adequate quality.

Please, do not cite the "reviews", just read products specifications, and measure them, if you wish.

As I said before, I will report dropped frame on D8 tape immediately after I can find one. Strange enough, 6 months after purchasing D8, I could not do this. Wait more, please.

Regards,

Grigory

[DrMordrid]

I know this one will likely draw fire ;-)

These images of the B&W EIA test pattern pretty much sum up what Camcorder & Computer Video was talking about. I'm keeping the comparison between Sony products so brand issues don't apply.

Sony TRV900 DV cam:



Sony TR7000 D8 cam:



Looking at the fine vertical wedge you'll see the DV cam can easily achieve over 500 lines while the D8 stops at about 330 lines on the coarse wedge.

In the horizontal wedges the DV displays about an 80-100 line advantage over the D8.

Also note the yellowish cast in the D8's image.

Yes the D8 is cheaper and the TRV900 is a higher end cam, but where is that DV quality 500 line resolution Sony is advertising?

It looks like this spec only applies to the theoretical abilities of the codec and not to what the D8 cams optics & CCD can deliver.

The price issue seems be disappearing. At the local Best Buy there are now several models of DV cams that are within $100-200 of the D8's.

Dr. Mordrid



[This message has been edited by DrMordrid (edited 10 November 1999).]

[Patrick]

Well, let's see their test results, Doc.

[Brian Ellis]

Doc!, I like it.

I should like to add something about lenses. A good single focal length still camera lens will resolve 400-500 lines/mm (the theoretical maximum for white light is determined by the wavelength of the highest frequency light the film/optics combination can 'see', typically about 1000 for violet). Assuming the film can resolve the same (very few can), a 500 lines/mm resolution will give a theoretical 864,000,000 picture elements in a 24 x 36 mm frame. The same lens on a 1/4" CCD, with a frame of 3.80 x 5.07 would resolve only about 19,250,000 elements. This is still impressive, but we are talking here of a single focal lens which costs, alone, the price of a cheap camcorder. A *good* zoom lens with a 3:1 range for a still camera will resolve typically about 150-300 lines/mm, varying with the focal length and, again, we are talking about something which would be prohibitively expensive for a consumer camcorder, albeit cheaper than a high-tech fixed focal lens. I doubt very much whether it is possible to design and produce a cheap zoom lens with a 12:1 range and a full aperture greater than f2 which could resolve 500 lines over a distance of 5.07 mm, i.e. c. 100 lines/mm, over the full zoom range, especially bearing in mind that, for lightness, some of the internal elements are in synthetic polymers which are relatively unstable with temperature and humidity, compared with optical glasses. I'm pretty sure that the camera makers are working to a limitative compromise of price, weight, materials, manufacturing techniques and performance. In other words, a high end camera will be better than a low end one, as Doc's illustrations show very eloquently.

This will also be shared with the electronics. A CCD gives quite a small signal output. To amplify this to a useable level requires a low-noise first stage in the amplifier. One way of apparently reducing the noise is to limit the bandwidth. If the lens is capable of resolving only 300 equivalent lines, as opposed to 500, in a high-end camera, it is possible to use a much cheaper front end to the amplifier and to add in a very cheap R-C filter to reduce the bandwidth and, hence, noise. There are probably tens of cost-cutting techniques used in cheap camcorders, each one reducing the image quality a tad.

I agree, this applies equally to cheap DVs as well as D8s.

In other words, "you pays your money and you gets what you pay for."

------------------
Brian (the terrible)

[Brian Ellis]

Grigory

I don't think you have quite seized on what I was saying. I agree that the VHS/TV set combination is not as good as one would like but the point I'm trying to make is that each compromise affects the final result. If you have a TV set with a resolution of 300 equivalent lines, you will still get a better picture if you feed a signal into it from a 500 line resolution camera than you would from a 350 or 300 line camera. The difference will be small but nevertheless easily discernible to even an untrained eye. The moral of the story is to keep the resolution at all stages as high as possible to have the best end result (cf. the Ilford tests I mentioned a couple of posts ago).


------------------
Brian (the terrible)

PS This is also why a DV image taken into a computer via Marvel is **considerably** better than one from any analogue camera. I'm unfortunately in this position because I bought an early DV (Panasonic NV-DX1) which does not have any digital I/O in any shape or form, so I have to use the analogue. I can assume that the D/A converter in this camera is good. I've not measured the resolution, but my guess is that it is considerably better in analogue O/P than any of the VHS/S-VHS machines I've used in the past.

[This message has been edited by Brian Ellis (edited 10 November 1999).]

[DrMordrid]

As I noted in my posting there is somewhat of a difference in price between the TRV900 and the TR7000. One thing to take into account however is that many of the newer/inexpensive miniDV cams can provide resolutions within 40-50 lines of the TRV900 results. This still puts them hell-and-gone better than D8.

My Panasonic PV-DV710 provides images that are in the 480-500 line territory and it only cost $750, below the price of some D8's at the time. Some mailorder stores are now selling it for $699. So much for the D8 cost advantage.

My main point was to illustrate that the D8's are not living up to hype. I think a lot of folks are falling for the advertising and Sony name in the absense of published visual comparisons. Basically many reviewers who will make these criticisms on the web or forums are not willing to take on Sony in print media.

As for the original article I DID read it and yes he likes the TR7000 for what it can do, but that is a subjective measurement. The images are not.

Does his willingness to accept the TR-7000's limitations for its use as a backup cam excuse the fact that it can't live up to Sony's advertising as to it's lines of resolution, or to even come close? That it seems to have a problem with its auto white balance? I don't think so.

The situation is that D8's are low and midrange Hi8 cams with few additions other than DV encoding and Firewire output. They use the same CCD's, lenses and tape mechanisms. If someone is willing to spend almost the same for this conglomeration instead of for a miniDV in the same price range then I guess that's that.

Dr. Mordrid



[This message has been edited by DrMordrid (edited 10 November 1999).]

[A_BIT]

Doc Mordrid,

It seems to me that the test results you posted are not useful in the context presented because they compare a 3ccd unit to a 1ccd D8 at 2.5 to 3X the cost even at mail order prices.

From what I have learned reading various articles on the subject, the color bleeding shown in the D8 test pattern you posted is common to all 1ccd cameras. If that were not the case there would not be a need for 3ccd units, right?

The 500 lines capability advertised for D8s clearly states that this can be acheived via tape dubbing only, not through the lens. This is typical of any digital camera in the price range.

Just my amateur $0.02...

-Anthony

[Grigory]

Hi,

Well done, Doc, but you forgot to read the original article with these pictures:
http://www.bealecorner.com/trv900/respat/

It is always good to refer to original material, as you probably know, as well as read everything down to the end of story.

If you read this article, then see the latest remark about the camcorder that I own:
http://members.carinthia.com/gpfeiffer/resolution.jpg

Do you see the difference in picture quality?

Or, poor americans, you were fooled by Sony!

BTW, I made my own tests with my own TRV110e, using the test image from this article, in April.

The picture was about the same as for the latest link. I also used analog Avermedia (same BT 848 chip as for this picture) card.

This was later reproduced with less analog noise on DV connection. See the explanations in the article.

This is another example how "limited" is the information given in semi-official journal publications, involved in competition.

Finally, I am pretty sure that we, happy europeans (in the far east corner of Europe) have 420 lines from optical shooting, and less chromaticity distortions than americans.

Not too bad for 1/3 price of 3CCD camcorder?

Terrible Brian ,
Yes, you are right with all words about optimizations, compromizes, else. However, the drop in 20% of luma resolution is a good optimization. As for most of users, do we have an ability to show these 540 lines on TV screen? What is the cost of TV with more than 720 light-emitting lines per screen?

Name, model number, ... , please! If it is not difficult, point to some source where I can see real-world test results for such TV.

What I have for reasonable priced units - 600 "pixels" per line on 29' Philips TV.

Well, S-video connection, other goodies,..., but who will produce TV set with honest 5 MHz output video amplifiers to display mainly 3 Mhz broadcasting pictures?
I tell you: only high end models can do this actually. Mainstream TV will show 300-330 lines on this test pattern.

Please show me where I am wrong.


Regards,

Grigory

[Brian Ellis]

Doc

Sure as hell, no magazine is going to be truthful about advertisers' products, if they are really bad or simply don't "live up to their hype". They may give some veiled hints that all is not well and we have seen this. It is a common journalistic trick, when something is bad, to bullshit with something easily seen to be wrong, and, if the advertiser complains when a review is bad, the editor can apologise and say, "Look at this bull: the guy who did this obviously did not know what he was writing about. Sorry, we'll print a correction in the next issue.", but it is rare for advertisers to do this because they think (poor sods!) that it is so blatantly obviously wrong that no correction is necessary. The devious minds of journalists.

However, I'll take you to task on CCDs. Sony Hi-8 CCDs usually have 270 or 320 kpixels, D-8 460 kpixels and Sony consumer DVs 680 kpixels. I believe the lenses on Hi-8s and D-8s are similar, but not strictly identical, as they are usually 20:1 on D-8s, against 18:1 on Hi-8s. This could probably make them worse: to have a 20:1 zoom must be a basket of compromises. Any lens designer will say that 3:1 is not too difficult, with a reasonable resolution; 6:1 is tricky; 10:1 is possible, with compromises which may be acceptable and above that you are in deep trouble, the trouble getting deeper as the ratio rises. You will rarely find more than 10:1 on pro or semi-pro camcorders, often less. These are designed for three points of near-optimal performance: at full wide angle, at full telephoto and at mid-range, as being the positions that cameras are used the most frequently, in practice. Everything in between is relatively poor. If you have a good camera, you can see this. Manually focus at full telephoto on a fine object about 10 m away and then zoom away from it. You will see visibly that the image will go slightly blurry, then sharp, then blurry again and then sharp again with many lenses.

Anthony:

Not quite right. Some 3-CCD cameras do have better luminance resolution as well (see my posting above) and better s/n ratio. You can find cameras with 500+ equivalent lines with only 270 kpixel CCDs (3) by this astuce.


------------------
Brian (the terrible)