How does Canon manage to destroy details?

Started by -sandro-, May 16, 2013, 07:03:17 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

-sandro-

Hello,
help me understand, we now can get 1080p RAW video from their cameras without putting extra work on the CPU, this means that there's an actual stream created by default by Canon's internal "code". So why in the world don't they use this data to compress videos but instead they do some strange line-skipping?



N/A

Because they're not video cameras by default, they're photography cameras that happen to record video as well.

If they didn't do line skipping, we'd be stuck with sensor crop videos or pictures at a max resolution of 1920x1080.

Also, these cameras are aimed at average consumers, who probably aren't too keen on recording massively sized raw videos whenever they want a few video files of their kids playing soccer, their friends getting married, yada yada.

7D. 600D. Rokinon 35 cine. Sigma 30 1.4
Audio and video recording/production, Random Photography
Want to help with the latest development but don't know how to compile?

1%

Its more complicated than just skipping... they aren't using the CPU to do much. More like semi-dedicated ICs inside the digic package.

-sandro-

I understand the product targeting part but I don't understand the technical part. Isn't RAW 1080p using the whole sensor? What does have it do with pictures or cropping? If yes wouldn't be easier for them to compress the RAW video we now have access to? It's already there !! or maybe I didn't understand the process completely...the RAW stream is something ML developers were able to "force" out of the camera?

qsara

it kind a is. but wait, many gpu and cpu makers lock their cpu for stable speed, ML not actually overclocking cpu, but increases its workflow.. cameras already designed for work at least 10 years, and in 10 years they'll be like film-cameras, so you can use ML, if you like it.

RenatoPhoto

Quote from: 1% on May 16, 2013, 07:20:24 PM
Its more complicated than just skipping... they aren't using the CPU to do much. More like semi-dedicated ICs inside the digic package.
Could it possibly be that Canon did not think that Sensor to DNG to Video pipeline was doable?  So the only way to get video is Sensor to Compressor to Video?  After all ML used CHKD engine to get DNG?  This question probably shows my ignorance on this matters, sorry!
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

Intenditore

Quote from: N/A on May 16, 2013, 07:13:41 PM
Also, these cameras are aimed at average consumers, who probably aren't too keen on recording massively sized raw videos whenever they want a few video files of their kids playing soccer, their friends getting married, yada yada.
Totally disagree. Dslr video is used by professional videomakers mostly, so Canon must think about. They are just too greedy. Take a look at 1D C - motion jpeg codec. It's quite ridiculous! In camera for about 9k$ there's not raw video but it's definately capable of. It's not a technical problem, it's managment.

So, I want to ask developers, is this possible to change canon's compression to something better, cleaner, with finer details and dr? Not raw, but something like ProRes?

g3gg0

as 1% already said, the hardware is doing all stuff.
sensor readout, downconversion, corrections, compression, etc.

we can only save what the hardware gives us.
the cpu is much too slow. it cant even memcpy the raw data in fullhd without problems.
Help us with datasheets - Help us with register dumps
magic lantern: 1Magic9991E1eWbGvrsx186GovYCXFbppY, server expenses: [email protected]
ONLY donate for things we have done, not for things you expect!