Bit rate investigation

[hjfilmspeed]

At this point canon is starting to make me sad. Have you seen the s35 1080p all-i from the 1dc. its tack sharp!

http://www.eoshd.com/content/9548/bbc-freelance-cameraman-tests-super-35mm-1080p-on-the-canon-1d-c

I feel like if the 1dc is shooting s34 4k with dual digi 5s why cant we get sharper s35 1080p from the single digi 5 cams like the 5d3.  I would so take the s35 crop if it ment sharper 1080. Sorry i know ML cant change the deep deep proporties of canon dslrs but i feel the need to voice this one anyway. I LOVE YOU ML! but i strongly dislike canon (coming from a 5d3 owner)   

[RenatoPhoto]

I LOVE YOU ML! but i strongly dislike canon (coming from a 5d3 owner)

Me too!

[wenen]

Me too!

Me too

[DTSET123]

me too

[1%]

i don't dislike how they made the digicV encoder... I do dislike that they didn't put lower tier features into a more expensive camera. ie crop mode.

I have to try this moire out that everyone is talking about. Time to get out the striped shit because its not doing it to bricks.

[3pointedit]

"Time to get out the striped shit..."

Wow you are dedicated, I just use a shirt!

[clkvang]

Is there a current release of this firmware for me to test out? Kinda interested to see what the results are for real world video versus the still examples posted on this thread.

Saw a few links on downloading, but not sure which was the current firmware everyone was testing.

[1%]

For 600D - top is newest
https://bitbucket.org/OtherOnePercent/tragic-lantern/downloads

6D  - top is newest
https://bitbucket.org/OtherOnePercent/tragic-lantern-6d/downloads

for ml stock just do the nightlies

[Kabuto1138]

Hey guys,

So is this dead?  what happened to Driftwood, i know he was helping out with the 5d3 but haven't heard from him a while.

[1%]

Not dead just H264 is at max on 600D. I could use help from driftwood on figuring out how to play back the 4x4 transform videos, I can't decode them.

Next step is MJPEG with custom buffer sizes to stop moire, etc.

[wenen]

Not dead just H264 is at max on 600D. I could use help from driftwood on figuring out how to play back the 4x4 transform videos, I can't decode them.

Next step is MJPEG with custom buffer sizes to stop moire, etc.

Good job! work on

Btw what is EOSMovieFixer.exe?

[1%]

For fixing atoms on movies over 4gb

[Kabuto1138]

Thanks 1%.  Mjpeg sound pretty awesome. 

All other cameras seem to be getting better Bit Rates through hacks except the Canons.  But Mjpeg would GREAT!

[Rush]

All other cameras seem to be getting better Bit Rates through hacks except the Canons.  But Mjpeg would GREAT!

Why "except Canon"? Thanks to ML developers we have control over Canon bitrate and we can control GopSize too! It is awesome.

[1%]

Well GH2/GH1 - software encoder so easier to change everything.
Nikon - No hacks that I know of, like 24mbit video
Sony - Has some hacks for like 42mbit

I'd say next to GH2 we are pretty good, only held back by the canon resizing.

[Kabuto1138]

My apologies,

Didn't mean to saying anything negative about ML.  I don't I would still hace a Canon camera if it wasn't for ML.   

I thought they had bitrate hacks for the nex?  maybe not. 

When I try to increase the bitrate with ML 2.3 it barely does anything, maybe 10 mbs extra? if I'm lucky.  Sometimes recording stops, even with a fast card, I rather keep it at 1.4x just to be safe. 

But I think Mjpeg would amazing.

[Rush]

When I try to increase the bitrate with ML 2.3 it barely does anything, maybe 10 mbs extra? if I'm lucky.  Sometimes recording stops, even with a fast card, I rather keep it at 1.4x just to be safe. 

ML unlocks upper limit for bitrate - it can be at 150 mbits and more. But it have some limits - bad buffer work (g3ggo have some workarounds; turning off audio will help too).
For me, at stock GOP, after turning off sound recording I can squeeze up to 80-100 mbps video (twice more than stock!) to get rid most of compression artefacts which is notable with high ISO.

I am sure you will see the difference: (upscaled 400% crop)

[1%]

When I try to increase the bitrate with ML 2.3 it barely does anything, maybe 10 mbs extra?

That way only tricks the prediction into predicting larger frames. Then it holds all of them (at gop12) in the buffer and stops. Mine adjusts lower level quality setting and sets it up or down based on buffer usage. You have to make it make bigger frames AND write them out faster.

At some point a scene won't compress any higher. Stock is really conservative though, it had to to work on < class 10 cards.

[Kabuto1138]

Hey Rush,

I do see the difference.  What the heck am I doing wrong.  What settings are you using for this?

thanks

[Rush]

What settings are you using for this?

It is not that simple with bitrate control.
First, if you record with audio - you will not be able to increase bitrate more than extra 15-20 mbps (1.3x).
But if you record without audio - you can switch to 2.0x on Class10 SD and up to 3.0x on speedy CF cards I think... (g3gg0 experimental tweaks can help with buffer overflow, but I didn't tried it by myself)

What will bitrate do to your video quality? For real, stock 45 mbps is enough bitrate for static camera scenes without a lot of movement or tiny details and when set to low ISOs - you will not see any downsides or artefacts to image cause of lack of bitrate.
But filming complex scene with movement + high ISOs (noise makes a lot of tiny movement across the frame) - requires high bitrates. 45 mbps become not enough and tiny details became washed out and blurred. It became more obvious when you start color grading and sharpening in post - you will see ugly lines and even blocks showing up as you sharpen your image.
High bitrate with ML can preserve more details on such bitrate demanding conditions.

So, the conclusion:
Stock 1.0x bitrate is usually artefact-free on low ISOs - 160 or 100 or 80 and trying to raise up bitrate will do nothing - it tops at 40-50 mbps. As you go up with ISO, you need more bitrate to avoid ugly compression lines and blocks which becomes visible after post shaprening.

Side effect - high bitrates on high ISOs will preserve not only fine details, but fine noise too - so it will look noisier that usual. But it will be easier to get rid of noise in post (NeatVideo) because it is less blotched or blurred.

You can clearly see if your compression works best and it have enough bitrate - just turn on bitrate info on screen and look at QScale value. Top bitrate quality is Q-16, and as it goes to 0 and to +... it means that your image is suffer of lack of bitrate.

[Kabuto1138]

Thanks Rush!

[Yoshiyuki Blade]

Yeah, recording without audio is the key factor in increasing video bitrate tremendously. As a matter of fact, when I fool around with camera settings and record indoors, I can just keep the quality at constant Q-16 and it won't go anywhere near my card's limit (~120 Mbps). It does exceed that limit outdoors though. There's just too much high frequency detail to maintain Q-16 with my card.

With audio on, I can't really exceed 60 Mbps. But with audio off, indoor bitrates typically range from 70-90 Mbps so it makes a sizable difference. With this much bitrate at our disposal, I wonder how much more quality can be achieved with mjpeg. Could it exceed what Q-16 of the H.264 encoder can do at similar/lower bitrates? I'm eager to see how things pan out.

[1%]

Q16 isnt really the tops. They are made up canon numbers. The real scale is a little bit bigger. Look at the 422 files for what input looks like. Mjpeg should look like that. 422 color is the main benefit.

600D can do the wav separately and 6D seems to just record audio with whatever bit rate you throw at it.

[wenen]

Some good information here about stuffs http://www.youtube.com/watch?feature=player_embedded&v=Fs1TnHjGljw

[Yoshiyuki Blade]

One thing about what the guy in the video mentioned that bothered me is that he somehow made a connection between a a bayer filter and the various types of chroma subsampling of the YUV color space (technically YCbCr). I noticed someone else make that connection in another video too so I don't know whether I'm misunderstanding something,  whether they're oversimplifying things or even misunderstanding it themselves.

Afaik, Bayer filters capture individual RGB colors and interpolate them since each pixel doesn't have all 3 color channels dedicated to it. Well, it's also called RGBG or some different order depending on the pattern because there are 2 green pixels for every 1 red and 1 blue in a 2x2 pixel block. Anyway, the raw image from the sensor can be "demosaiced" internally or via a RAW photo editor and can be encoded in RGB. I don't see how this has anything to do with 4:4:4, 4:2:2 and 4:2:0; that's from a completely different color space. Unlike RGB, where each pixel has its own RGB channel, each with its own luma level, YUV has 1 channel dedicated to luma and 2 channels dedicated to chroma, which makes it much more efficient. It also allows images to have a lower chroma resolution than luma, exploiting the human eye's weaker sensitivity to chroma resolution over luma resolution. Something like that doesn't happen in RGB. They're also different enough that they can't be losslessly converted between one another. A full 8-bit RGB picture can be converted to YUV 4:4:4 and back and look very close, but there will be rounding errors.