Does this bother anyone?
I was just wondering if anyone noticed or is bothered by the last shot of the now unstickified, "Extensive HD100 / Mini35 Hands-On Test" by Charles Papert, "Night walk left." http://www.hdvinfo.net/articles/jvcp...ini35test2.php
I haven't noticed the SSE which some people have spoken of, but I have seen trails behind the actress' arms & body. This is the only clip I've downloaded from the HD-100 that shows it this bad. Here's a frame grab: http://video.terpstar.com/extra/JVC-HD-100_Capture.jpg It may not seem like much, but I noticed it on the first play through and was surprised when I went back and saw how bad the trails/ghosting was. Like I said, every other clip from this camera is not like this at all. What do you suppose happened? Early version of the codec? Has anyone else had footage do this? |
I should point out that this was a pre-release version of the camera.
|
I haven't noticed anything like this on the released versions. I did see it once on the very first HD100 I had while it was in DV mode. They fixed that right away with a firmware update.
|
Thanks, guys. I admit it made me nervous to see. I feel much better now.
|
OK, this is probably just my eye after doing miles of composite film but is that a VR background droped in behind a walk-by shot with an HD100 or was that an actual exterior with no chroma?
|
It's a legit shot, not composited in any way, just shot at night. That's from when we shot with the HD100/mini35 combo. A few months ago I was going through that footage and was alarmed to see that kind of streaking in a lot of the shots. I'd never noticed it before.
As Charles said, that was a pre-release camera. I don't know if the effect was due to it being a pre-release issue, or if the mini35 somehow compounded the effect. It only happened in the dark shots, I never saw evidence of it in the daylight shots. To be fair, some of those shots Charles designed really pushed MPEG about as far as it could go -- for example, in that particular shot, that was a fast dolly-in, which is about the ultimate nightmare scenario for MPEG motion prediction -- everything grew in size in the frame quickly, and MPEG can't predict motion like that. Combine that with the seemingly-random texture of the vibrating ground glass of the mini35, and I could see how MPEG could end up choking. And I've seen a limited version of a similar trailing artifact in Nate's downtown clips (a trailing shadow chunk of a building during a pan), so I'm pretty sure that's what it is. It also absolutely didn't happen on the live feed, so I think it's due to the mpeg compression choking on the moving ground glass texture. There were many shots where there were multi-frame trails, all in the night footage. It would be illustrative to get a production HD100/mini35 combo and try it again to see if that's something that was limited to the pre-release model; I believe Stephen Noe said something about a ghosting issue in the pre-release version that was cleared up before production models hit the street in the US. |
Quote:
It wasn't as bad, but certain parts of the image would smear slightly almost as if the grain pattern would not refresh as you would expect or if there was a static GG in front of the lens. I haven't seen much evidence of this elswhere on the numerous clips I've downloaded. (Quite a few nice ones here by Michael Pappas: http://www.gyhduser.com/showthread.php?t=359) It's also interesting to note that you can't see it at all when converted to SD and played back on a CRT. At least not with Nate's clips. I think he also did a shoot in November with the Mini35 combo. Here he talks about a lot of grain in the shot, but I don't know if that's the same smear effect we're talking about here. http://www.dvinfo.net/conf/showthread.php?t=55111 I would be curious to see that footage full-frame. |
I agree with Barry that the original night shot we're discussing does stress the codec in some crazy ways.
Somebody did point this effect out on some later samples I posted, as Barry mentions above also. I did notice it after pointed out...did not notice it prior and haven't since on anything else I've done. The music video from last Nov doesn't have any that I've noticed either. When I talked about grain problems with the Mini, I was talking about loss of detail because the codec couldn't render all the grain correctly. |
Quote:
You can see it in this still (1:1 scale, cropped): http://ethur.stargate.ca/neopics/hd1...ement_echo.jpg You can see a distinct outline of where the person was in the previous frame. It's quite noticeable during playback (even moreso than the still because it changes between frames), but not overly objectionable. I'm guessing it's due to the MPEG2 compression -- these are probably B or P frames. |
I should also add -- my sample is just with the HD100 and stock lens, no Mini35 adapter or similar device.
|
Earl, I'm having trouble loading the file. Could you upload it again maybe?
|
Here's another link to the file on a different server:
http://www.neopics.com/images/hd100_movement_echo.jpg If anyone has a problem with either link, it would be greatly appreciated if someone could send me a screen shot of what they see in their browser. I have no problems seeing these at home or at work, so I need something I can show my tech guys. |
It works for me now. Thanks Earl.
I do see the echo very slightly, but it's not really bad. It seems more like the grain pattern decided to make an outline of her. I probably wouldn’t have seen it on this still if I wasn’t looking for it. But these small imperfections can get to me in playback if there are enough of them. -just a tiny reminder that you're watching a reproduction of an event and not the event itself (it can take you out of the moment). |
All times are GMT -6. The time now is 10:59 AM. |
DV Info Net -- Real Names, Real People, Real Info!
1998-2024 The Digital Video Information Network