Film to Mindv transfer question

Post by **crimsonson** » Wed Feb 05, 2003 11:12 pm

Its probably doing some advance filtering. With such a high end card card it would be hard to really distinguish the difference using a standard television. A better test is to output a uncompressed Bitmap file for both versions and check it in Photoshop's Histogram. You will be able to see noise, color banding and such of each version.

DV Magazine tested it once a couple of years ago. Y/C output of a DVCam scored slightly less than DV and this is in all categories. It was across the board. It was not a huge difference individually, but collectively DV did score high enough that I am sure people who are trying to squeeze out every bit of quality from video [like us cineastes] would definetly pick DV.

Post by **MovieStuff** » Thu Feb 06, 2003 4:22 am

crimsonson wrote:With such a high end card card it would be hard to really distinguish the difference using a standard television.

But isn't that what most people watch video on? And, really, the difference is pretty significant, at least on my monitors. The signal directly from the SVHS port looks much better than the compressed signal via DV. I know you don't believe me and I understand your reasoning but you really should try it and see. On my system, at least, the difference is quite noticable. I still wonder if other miniDV cameras pick off the SVHS signal before compression or if any decompressed the DV for output as SVHS.

crimsonson wrote:A better test is to output a uncompressed Bitmap file for both versions and check it in Photoshop's Histogram. You will be able to see noise, color banding and such of each version.

I'll be sure to tell all my clients to do that after I send them their transfers. ;)

Roger

Post by **S8 Booster** » Thu Feb 06, 2003 9:56 am

Not sure if this is valid for all cams.
Seems like the S-Video tap is ahead of the compression processing stages.

R

Post by **mattias** » Fri Feb 07, 2003 3:56 pm

S8 Booster wrote: Seems like the S-Video tap is ahead of the compression processing stages.

does it really seem that way? to me it seems that this scheme doesn't include the svideo *out* at all, and that the rgb signal from the camera is fed directly to the adc, which makes sense. if this is the case you'll never get anything better than the digital format used (720x480/576x4:1:1/4:2:0) whereever you tap off the signal, but you should theoretically be able to get it before the dct which would be nice. do you have any documents on the output part of the camera?

/matt

Post by **MovieStuff** » Fri Feb 07, 2003 5:20 pm

All I know is that coming off the SVHS output the picture looks very smooth and artifact free. Coming out the firewire looks blocky. If I had to choose between 500 lines of blocky imagery or 400 lines of smooth imagery, I'd choose the smooth imagery any day of the week. There is certainly more to "a good picture" than sheer resolution or none of us would be working in Super 8. ;)

Roger

Post by **mattias** » Fri Feb 07, 2003 5:50 pm

MovieStuff wrote:Coming out the firewire looks blocky.

then there must be something wrong with your setup. the compression artifacts shouldn't be visible at all in most cases, and when they do show they just look like video noise and/or color bleeding -- never blocks.

/matt

Post by **MovieStuff** » Fri Feb 07, 2003 7:33 pm

Okay, we'll use the term "noisey" instead of blocky. Whatever. There's nothing wrong with my set up. I've seen the difference on every miniDV camera I've worked with. When we take the analog out the image looks smooth. When we take the firewire out, the image looks not as smooth and definately more "noisey" or "grainy". Blocky was probably not the best term but it definately looks compressed in some way compared to the SVHS output. Anyway, technical issues aside, I think the SVHS out looks better and when I use it with my Canopus convertor, the firewire out on it looks much better than the "built in" firewire converter on the camera.

Post by **mattias** » Fri Feb 07, 2003 7:39 pm

MovieStuff wrote:it definately looks compressed in some way compared to the SVHS output.

ok. :-) one question though: what about when you play a tape from the deck/camera directly to the video monitor via s-video? perhaps the dac in the camera is just better than the one in your computer or something?

(i did notice your remark about the canopus fw out looking better than the camera's, but i quite don't understand what that means. how does firewire look? it's just a digital signal. you still have to convert this to analog video to view it on the tv, right?)

/matt

Post by **S8 Booster** » Sat Feb 08, 2003 8:53 am

This scheme being a simplified model of DV processing for recording. From a technical viewpoint it would be logical to simply reroute the signal back to the S-video port after the initial processing (4:2:2/4:2:0) prior to all the following DV processing stages on its way to the media.

Interesting that the problems in motion for DV are the same today as they were 10 years ago.

You can check the complete site here:
http://www.oakmusic.com/parkplace/Video ... _formt.htm
http://www.oakmusic.com/parkplace/Video/TechPapers.htm

The cameras mentioned as principaly identcal to above is DCR-VX1000 and DHR-1000.

I did some very simple test with my new Sony DCR120 by connecting it to my old MAC 840 AV from 1994 using the S-Video connection and the MACs built in AD converter which can be set up to 1024 pix width on that old machine! Simoultaneously I recorded to the DV tape in cam. I will post samples later but my initial impression is that both Mattias and Roger are right.

Captured via a HE capture card/S-video gives better STILL images than DV stills.
(uncompressed PICT format - HQ JPGs should be identical)
For moving images there are no visual difference to me.

It seems to me that the DV processing sort of doubles the image quality when the film is in motion compared to the stills. Can be something with the DV codec? (Dynamic processing)
When a film is run on QT6 the stills can appear grainy while the moving images are not-

Post samples later.

R

Post by **S8 Booster** » Sat Feb 08, 2003 9:18 am

Posting images in PICT format as captured to not mess with any comression equalizers.

The following images are from a rip off setup not verified but can be seen as an indicator.

Original from DV Tape: 768x576: 141k
ftp://ftp.filmshooting.com/upload/pictu ... 8x576.pict

Captured from the camera head via S-video to the MACÃ‚Â´s AD at no compression: 768x576: 988k
ftp://ftp.filmshooting.com/upload/pictures/768x576.PICT

Another capture from the MAC sized to 832x624: 1122k
ftp://ftp.filmshooting.com/upload/pictures/832x624.PICT

For reference only from the MAC 729x576: 848k
ftp://ftp.filmshooting.com/upload/pictures/720x576.PICT

The standard 1994 model MAC digitzer should not be up to modern capture cards but I think still it comes out surprisingly well compared to the in cam AD converter. Can capture up to 1024 pix width.

PICT is an image pict format common with MACs and will normally not open in default browsers. Normally got to be downloaded and viewed in an image viewing pogram. I am certain there are image viewers on PCs that cope with them as well but I do not know which. Guess Photoshop is a good bet for starters.

R

Post by **S8 Booster** » Sat Feb 08, 2003 12:27 pm

Details from the pix above. "Uniformed" at 3072 pix witdh prior to cut detail:

Original DV 768x576 img:

MAC AV 768x576 img:

MAC AV 720x576 img

MAC AV 832x624

R

Post by **S8 Booster** » Mon Feb 10, 2003 11:56 am

Regarding viewing the PICT files above.
If you download the free Quicktime Player for PCs a Picture Viewer is added in this package. PV opens the PICT files.

http://www.apple.com/quicktime/products/qt/
http://www.apple.com/quicktime/download/

Regarding the results the DV images look better on my Laptop PC while the MAC/Premiere/S-video images looks better on the MAC so this is elastics by the meter to me. MAC results are always less "squarified."

Apart from this I am confident that if the S-video connection is set up right with a high-end spec cable to reduce the contour effects it will easily match or surpass the resolution of the true DV results.

A more professional setup and more correct amount of light - this was pretty low, would probably improved all results.

R

Post by **mattias** » Mon Feb 10, 2003 1:53 pm

sorry to sound like a broken record, but you are viewing the dv footage at "high quality", right? what results do you get if you play back a dv tape and capture analog?

/matt

Post by **S8 Booster** » Mon Feb 10, 2003 2:36 pm

No problem Matt.
I really fancied the old (superb) linear tracking recodo pureas, they did not hook up in the scratch so easily.

I will make a try on the setup you describe. Only problem is that the old 40mHz MAC struggles with 768x576 captures as the frame rate drops at motion in picture. Anyway, due to the similarty with MiniDV you can actually see the codec "work" as it will capture and reuse identical frames when the cam or object is not moving so i believe I can work it out. Allow a little time.

R

Post by **S8 Booster** » Mon Feb 10, 2003 4:50 pm

Sony DCR PC120 original recording - In cam DV tape -> S-video playback via MAC 840 A/D converter format 768x576. Still export PICT format.

Frame grab: ftp://ftp.filmshooting.com/upload/pictu ... 8x576.PICT

Sample mag to 3072pix - pix interpolation.

Well,......?

R?