Testing video quality override - General Help and Assistance on using CHDK stable releases - CHDK Forum

Testing video quality override

  • 1 Replies

Offline reyalp

  • ******
  • 12586
Testing video quality override
« on: 08 / January / 2011, 18:30:07 »
From http://chdk.setepontos.com/index.php?topic=5889.15

All tests using D10 and CHDK trunk rev 1031.

Each test was ~5 sec holding (fairly) still and ~5sec panning.

Full test files http://www.zshare.net/download/84943210a8c8361b/

Some frame grabs for those who don't want to download ~75 megs http://www.zshare.net/download/8494333574e32e8b/

DEF - quality set to "default", no override
QNN - quality override value.

Frames grabbed with avimux as bmp, converted to png with irfanview. Frame 195 was grabbed form each video (from the panning segment)

Lighting was poor, overcast through window + incandescent. Shutter speed may have impacted panning shots slightly, but by comparison to the still shots doesn't seem too significant.

- Low quality override definitely works. File size is greatly reduced, at the expense of image quality.
- Difference between higher quality and default is marginal. I think I can see some difference between 98 and default looking at the frame grabs, but even if it's real it's very subtle. I don't see it have any value for most uses, at least not on this kind of scene.
- Based on file size, default doesn't appear to be 84.

Don't forget what the H stands for.

Re: Testing video quality override
« Reply #1 on: 08 / January / 2011, 19:53:14 »
Thanks, that is useful.

A number of recent posts had suggested lower-quality override could be useful.

Although the upper quality limit could be hard-coded I guess there is not much point.

If people want to use it, that is up to them ...


Looking at the low quality frame, I see a patchy pink tint.

Exactly the same as on a video that someone sent me (from a cheap camera) as the camera panned.

I would guess the patchiness on his video was due to the extra compression that panning detail needs if bitrate is constant.

The shots were with tungsten halogen and I would guess the IR-block filter is simply not efficient enough.

« Last Edit: 08 / January / 2011, 20:05:32 by Microfunguy »


Related Topics