New workaround for the quicktime h264 problem?

[Holger Neuhaeuser]

Hi
I´m just curious if anybody found a new workaround for the problem of crushing blacks and whites (Rgb values 16-238) when importinng the h264 quicktime into editing programs like finalcut.?

My workaround is capturing the hdmi signal , it seems to be the only way to get it done at the moment.

Happy about any new suggestions.

Holger

[Bill Ravens]

not that I've heard of
the problem lies with the fundamental philosophy applied by the QT programmers. I don't think a single one of them uses a camera. This is a very old problem that persists thru many versions. when someone found a fix, they quickly disabled it. what does that say about QT amd Apple, in general?

[James Miller]

Holger,

As you are on a Mac the workflow to avoid/control the issue is.
1/ Edit sequence in native H.264.
2/ Send sequence to 'Color'. a) Amend the Luma curves to pull back the values. b) Colour grade.
3/ Render and send back to FCP.

Note: Color will render your footage in ProRes. Your new sequence in FCP will be in ProRes.

4/ Export sequence when finished, still in ProRes.
5/ Render to desired codec with 'Compressor'.

Note: If you machine will not play back H.264 smooth enough to edit with, you will have to pull the clips directly into 'Color' adjust and render/export first.

James

[Holger Neuhaeuser]

Thanks for the replies,

hm pulling back the curves in "Color" shurely stretches the signal back to the full range,
but are you shure that you haven´t lost Detail in blacks and whites already?

I think I´ll stick to capturing thru an hdmi capturing card (like Blackmagic intensity), cause it gives me the full range with no problems.

Only real drawback is that you have to capture it in "realtime".
But I came to the conclusion that converting to prores takes the same amount of time.
I doubt that the method you lined out with "color" is much faster.

Again, thanks a lot for the tips, but I hoped there was a solution that really got the quicktimes to work without regrading.

Regards
Holger

[Don Miller]

Holger, your capturing the 1080i signal out of the hdmi port that's available before you press record?

[Holger Neuhaeuser]

Originally Posted by Don Miller

Holger, your capturing the 1080i signal out of the hdmi port that's available before you press record?

yes, I tried two ways that basically worked with identically result:

a) converting the hdmi signal out of the 5d with a convergent-design convertorbox to HD-SDI, and took the resulting hd-sdi signal to the hd-sdi input of a blackmagic HD capturing card.
b) feeding the hdmi signal from the 5d directly into a blackmagic intensity hdmi capturing card (much cheaper)

In both cases I used the slideshow function of the 5d playing only movieclips, and captured the content of the whole card in one go.

I did this cause I realized how different the quicktime video looked compared to the playback out of the 5d, especially with night shots with lots of low black levels.
I found those shots nearly unusable from the quicktime file, while the captured version looked perfect.

[Holger Neuhaeuser]

Not to confuse anybody:

This works only with clips recorded to the cf card.
Not with the preview function that comes out of the 5ds hdmi port while shooting.

In other words, its just a working way of getting your recorded clips properly to an editing system, without further loss.

[Evan Donn]

Originally Posted by Holger Neuhaeuser

hm pulling back the curves in "Color" shurely stretches the signal back to the full range,
but are you shure that you haven´t lost Detail in blacks and whites already?

You aren't stretching the signal back to the full range, you're pulling the lost detail back from outside the 16-235 range - you can see the detail come back in both the video and the scopes. As long as you go back out to a 10bit format (like ProRes) you shouldn't lose any information.

Only real drawback is that you have to capture it in "realtime".
But I came to the conclusion that converting to prores takes the same amount of time.
I doubt that the method you lined out with "color" is much faster.

The Color workflow means you only convert the footage you end up using, so depending on your shooting ratio and total amount of footage it could be significantly faster. If you're going to color correct anyway the Color workflow doesn't add any additional time or steps to the normal post production process.

It also means you don't tie up your camera doing capture, so you can be offloading one card while you continue shooting with the next card, and someone can be editing a rough cut while you're still shooting. With the caputure method you either need to make sure you have enough cards to cover your entire shoot or you need to break to capture so you can clear your cards.

So in the end the best method may depend a lot on the nature of your shoot, as well as the time constraints you are working under.

[Holger Neuhaeuser]

I see your point Evan,
sounds good,
I´ll try that.

After all its a bit weird to do capturing like from a type based system,
while already having access to a nonlinear cardbased workflow.

Thanks
Holger

[Tyler Franco]

You can't solve this entire problem by using a picture profile like James Miller created to record a much "flatter" video in the first place?

[Don Miller]

What about just recording the HDMI output without pressing the record button?

[Josh Dahlberg]

Originally Posted by James Miller

Holger,

As you are on a Mac the workflow to avoid/control the issue is.
1/ Edit sequence in native H.264.
2/ Send sequence to 'Color'. a) Amend the Luma curves to pull back the values. b) Colour grade.
3/ Render and send back to FCP.

Note: Color will render your footage in ProRes. Your new sequence in FCP will be in ProRes.

4/ Export sequence when finished, still in ProRes.
5/ Render to desired codec with 'Compressor'.

Note: If you machine will not play back H.264 smooth enough to edit with, you will have to pull the clips directly into 'Color' adjust and render/export first.

James

James this really works, thank you. Just opening up the files in Color the low end magically appears. Wonderful. I'm still going to reduce contrast in camera and shoot flat as I think the camera does have a tendency to crush the blacks, but this together with the workflow you outline above my low end concerns are solved :-)

[Holger Neuhaeuser]

Originally Posted by Don Miller

What about just recording the HDMI output without pressing the record button?

I think it´s not the full hd resolution and you can´t get rid of the camera data that is displayed in the hdmi signal while in liveview

[Soren Hiorth]

hi

Thanks for all the great input in the thread.

I'd like to know, what the HDMI-output looks like, when the camera is not recording. I mean, is the data displayed on the screen placed in a way, so that you could crop it to 2.35:1. I find the h264-compression a little harsh, so would love to find some work-around, that gave a pure HDMI-output.

Have anyone heard of an alternative firmware that could help to solve this??

S

[Mike Calla]

Holger:

I didn't want to hijack your thread with frame rate issues so I started new thread:
http://www.dvinfo.net/conf/canon-eos...-problems.html


...but in fact i want to ask how did Intensity handle the frame rate? can you set it to capture a frame rate besides 30p