Understanding Conversion From MLV to Prores

Started by Pedr, January 10, 2015, 06:04:04 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Pedr

I've spent a lot of time today dredging the interweb in an attempt to understand the optimal way to convert from mlv to Prores. I understand how cumbersome cdng files are through my own experience and I understand the benefits of converting to Prores for editing, however some of the subtleties of this process are currently escaping me. If anyone could take a moment to put me right I'd much appreciate it.

I'm currently looking at going from MLRawViwer -> DNG Sequence -> AfterEffects -> Prores. I like this workflow because I know ACR (and Lightroom) very well, so I know how to get the RAW files into the shape I want them. My question is what shape should this be? My initial assumption was that I would grade the RAW to look the way I wanted in Aftereffects and then export, but heavy grading at this point seems very premature. So am I right in thinking that what I really want to do is to ensure that as much information as possible is extracted from the RAW files and present in the encoded Prores file. This would then allow more latitude in grading later on down the line. If this is a sensible approach, what should I do to make this so? How should I grade the RAW file to maximise the amount of information in the Prores file? As far as I can gather I absolutely have to get the white-balance and colour-balance right at this point, but what else should I do (if anything)?

chmee

mlvrawviewer is able to convert straight to prores, isnt it? and you nailed down the three important points:

(1) black and whitelevel
(2) neutral whitebalance => colortemperature/tint
(3) a curve aka lut to get most out of it

BUT (3) is the point you can overstate the case. a lot of curves/luts are havin' a "look". for a good workflow you should do the "look" while/after editing, but never while transcoding first footage.

regards chmee
[size=2]phreekz * blog * twitter[/size]

Pedr

Hi Chmee. Thankyou for for replying.

I realise that MLViewer can handle(2) white-balance/tint, but It only offers me control over exposure. There is no means to set (1) black and white level independently, so I don't see any other option than to use another app for this step. I am assuming that moving black and white levels independently to achieve the optimal/desired spread of values (as I would with a photo) is the right way to go, or is this also something that should be dealt with in a later grade? If this is so (and setting just exposure is enough), what should be my guide in setting the exposure?

As for(3), thanks for confirming that. I guess my difficulty here is that I understand the goal, but not the means. As you say, some of the LUTs are pretty severe. Is something like C-Log an appropriately neutral LUT to allow for flexibility in the grade later down the line?

DeafEyeJedi

Then try MLVFS and use AE... Maybe the way you want to use the ACR within AE to have more control?

Then export as ProRes from There or perhaps you can grade even more in DR and still be able to export as ProRes as well.

JM2C.
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

dmilligan

Black and white levels when talking about raw data are not something you adjust. They are either correct or incorrect. And yes there is a correct (mathematically speaking) value for both of these. This has to do with the linear nature of raw data (raw data is linear with an offset, black level is simply the value of this offset, that needs to be subtracted out before you do anything useful with the data). If you have a wrong black level, your image will have strong color casts, white balance will be impossible, and none of the color space transformations will work correctly. White level, while not as important is still either correct or incorrect. If it's too low whites will be needlessly clipped, if it's too high, specular (clipped) highlights may appear gray, or have color casts.

You may be confusing black and white level adjustment with "blacks" and "whites" sliders from ACR, this is a totally different thing.

Pedr

@dmilligan Thanks. You are bang on the money. I am confusing the two.

So assuming the footage is (reasonably) well exposed, if I set the correct whitepoint and choose a neutral LUT in MLRawViewer before exporting to Prores, this will give me that greatest flexibilty when grading the footage after it has been edited?

DanHaag

Cinelog is what you are looking for, I guess. It's available for Resolve and Adobe Camera Raw. You should check out their website: http://www.cinelogdcp.com/space-saving/ - There's also a thread about it on this message board, Cinelog's developer Andy is answering questions right here: http://www.magiclantern.fm/forum/index.php?topic=10151.0

Pedr

Thanks DanHaag. That does indeed look like a good solution.

reddeercity

I use A.E. with ACR to ProRes 4444XQ,  this the best option as All the other converter of ProRes are 10bit from FFmpeg
Where as the XQ flavor is 12bit or 16bit with a Alpha and rumored to be Linear in Color Space (Good enough for the Arri Camera's https://www.arri.de/news/news/alexa-xrxt-supports-new-prores-4444-xq/) ;D
The New ProRes Codec was designed from the beginning to be used with Heavy Grading and VFX
My 2 cents   

DeafEyeJedi

5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109

andresharambour


blainesuque

Does anyone know how to export our cinelog C Masters to prores 4444XQ from AE? i was able to get the option in the codec choices tab, but i get a error when rendering out 4444xq. Project settings are 32 bit float.  but i am on yosemite 10.10.2 which i dont think should make a problem but maybe?

I get the error Codec Compression Error

I took the component from FCPX and put into my quicktime folder , i also installed compressor 4 and it works in compressor but not in AE.

DeafEyeJedi

@blainesuque:

Please do not double post.
5D3.113 | 5D3.123 | EOSM.203 | 7D.203 | 70D.112 | 100D.101 | EOSM2.* | 50D.109