Stupid color corection question....

Chris Johnston

Chris Johnston

Veteran
The vast bulk of my stuff is shoot, and out the door, so I'm just trying to stay a little familiar with the tools I have. I'm crusin the interweb thingy and I find this color correction tut that I thought was just awesome. (basic but awesome none the less)

Color correction vid

The thought came into my head after watching...(and I don't know the difference, if there is one) but this tut shows how to correct your colors etc.
but what if I wanted some wierd varient like.... I want it to look like the Matrix (all green tinted) or all cold blue tinted like soo many of the sci fi movies nowadays.
Should I correct all the footy (like in the tut) and make it perfect. THEN mess it up it so it looks green like the matrix? is it more destructive doing it that way? (IQ wise)

If I want to make it a blueish/cold tint for the final product, should I be shooting with the WB shifted to the blue side (or use a picture profile that leans to the blue side) OR should I white balance correctly and then put some post on it to "cold it up" ?

(Hope I worded that so it's understandable????) I basically want to know....In the video, he shows how to match 2 shots of a hand on a piano. and he corrected the bluer hand to match the perfect skintone on the "red-er" hand. What if I wanted to have the final product colder? Would you "UN" correct the red-er hand to match up on the unperfect bluer hand? Is there any difference how the workflow goes? or any difference in IQ?
 
I always try to shoot as color neutral as possible but that's me.

Color correcting is extremely simple. Just take your color sampling tool put it on a white or black and measure the RGB value.
Then just push or pull R, G or B in curves until they all are equal. Simple and painless.
 
Chris Johnston said:
The vast bulk of my stuff is shoot, and out the door, so I'm just trying to stay a little familiar with the tools I have. I'm crusin the interweb thingy and I find this color correction tut that I thought was just awesome. (basic but awesome none the less)

Color correction vid

The thought came into my head after watching...(and I don't know the difference, if there is one) but this tut shows how to correct your colors etc.
but what if I wanted some wierd varient like.... I want it to look like the Matrix (all green tinted) or all cold blue tinted like soo many of the sci fi movies nowadays.
Should I correct all the footy (like in the tut) and make it perfect. THEN mess it up it so it looks green like the matrix? is it more destructive doing it that way? (IQ wise)

If I want to make it a blueish/cold tint for the final product, should I be shooting with the WB shifted to the blue side (or use a picture profile that leans to the blue side) OR should I white balance correctly and then put some post on it to "cold it up" ?

(Hope I worded that so it's understandable????) I basically want to know....In the video, he shows how to match 2 shots of a hand on a piano. and he corrected the bluer hand to match the perfect skintone on the "red-er" hand. What if I wanted to have the final product colder? Would you "UN" correct the red-er hand to match up on the unperfect bluer hand? Is there any difference how the workflow goes? or any difference in IQ?
Click to expand...

There's a difference between color correction and color grading.

Color correction is fixing a problem. Color grading is altering the overall color to give a distinct, intended feel.

The question for this comes down to thorough pre-preproduction and knowing what you want before you ever shoot.

It is far less taxing on the entire process to get the look the you want while you're shooting. Or, knowing exactly what you're going to do in post to get the look you want and shooting the footage in such a way to prep for it -- for example, shooting a "dark" scene with a lot of light to preserve detail and reduce noise, and then darkening it in post to achieve the final look you want. But, this requires knowing what you want before you shoot, not deciding after.

You certainly CAN shoot everything "flat" and then grade later, but are you doing that because you think it gives you a better result, or are you doing it because you don't want to decide until later what you want to do with the footage? I would strongly advise against the latter. "Preserving options for post" usually strikes me as not really knowing what you want to do, and kicking the can down the road to decide. You will get consistently better results with less work if you make those kinds of decisions before you shoot than you will making them after.
 
It's a totally valid question, not stupid.

The answer is not as simple though. If you're shooting with a compressed codec than it's probably a good idea to think about a baked in look. As any moderate grading will probably break the codec. But if you're grading isn't going to be very moderate then I'd just shoot as clean as possible. What would I consider a compressed codec? Anything below 50Mbs on a 25p frame rate I guess. Anything below this is a risk to moderately grade. For instance I just tried grading some 24Mbs AVCHD material form Canon C100. And you really have to be careful about what you're doing. The levels can't be pushed much. Don't get me wrong - you can grade the footage but you have to be careful. Shooting with the C100 and using a Ninja now that's another story. But this is rather subjective. You can find lots of 7D, 5D, GH2, C100, etc. footage that's been moderately graded and it looks good. But the codec is stressed and you can most probably see that. Will a consumer see that? No. It depend on what are your quality standards. But most people to my knowledge prefer to shoot neutral. It's a shame if you change you're mind in post about what you want to do with the image:) And it's hard to get the look you want while shooting. It takes time and effort which you probably won't have enough. Time is always of the essence! And plus it may not be what you thought you got until you'll see it in post on a nice big monitor.

As for the first balance the material and then grade it... If you first balance the material and then grade it there is going to be more degradation off-course. But this would depend on the workflow. For instance if you first balance the material and render it and then grade it sure, you're pushing the material twice instead of getting it right the first time. But you don't have to render the material after the balance you can go on without rendering. This way there won't be such a significant IQ loss. And there is off-course the question of the raw material quality. How and with what was it shot. If the original material has a good IQ and a strong acquisition codec was used then balance first. If not, have a second thought about what to do. Make a test, try grading a shoot directly and try grading the same shot after balancing first. Then compare the results and see if the IQ fall-of is significant if not then balance first. It's generally always good to balance first because you'll go nuts if you don't:):) It's a lot more objective if you balance first all you're shoots and then grade. You'll see once you'll have a longer project to grade (assuming you haven't done so already).
 
since your movie will certainly includes scene shot at different place and different time, you will need to color correct them to make sure everything
looks ok (for example that same color looks always the same across scenes) , then when the work is done (totally edited) , you will finally color grade the movie as needed. So yes, color correction and color grading is ok, and color grading will be easier if color correction is done properly.
color correction is technically required, while color grading is aestethically desired.
It is next to impossible to shoot a perfect picture reaching proper correction and grading at the same time, since often the two feature are just couteracting, or you need
to work at the good old way hollywood was, spending millions in lighting and costumes and painting full stage if needed.
 
Last edited:
Thanks for the advice guys. I should have said I am working with 2 fs100s, no external recorders, just the standard codec. Most of the stuff I currently shoot only requires minor tweaks in brightness, colors, and the bulk of my workflow is correcting audio. I have cineform, which I've used a few times, but to be honest, because I'm only doing minor tweaks I just didn't see any advantage in IQ, so I quit using it. (and saved myself the render time so I could get it out the door faster)
I am getting ready to shoot something that will be seeing much more in the way of image manipulation/hard grading so I'm thinking the best way to go would be to shoot, flip everything through cineform, then do the correction / grading. I know the extra render to cineform adds a render, but I think the 10 bit 422 will hold up better?? I'm kinda undecided on what is better (because I don't know...)

try to push the fs100 8bit 420
or
add the cineform render, then work with the "sorta" 10 bit 422. (sorta, because the 10 bit 422 is made from a 8 bit 420 so it's not really the same as an external recorder would get you, but probably better than the standard fs100 codec.)
 
Transcoding to a 10bit 422 codec is going to help down the road regarding IQ off-course. And also AVCH 24Mbs isn't a working post. codec, it's only an acquisition codec. Cineform is a postproduction codec that is meant to uphold IQ between renders (I think...haven't used it yet) and it'll be A LOT easier on the processor then AVCHD as it's an I-frame codec and it's not as compressed as the AVCHD.

First do a rough cutt with AVCHD then for offline transcode to cineform so you will have nice and smooth operation when editing your final cut. When you're ready for online (you've completed your edit) just use the same sequence and that's it. I use Avid so after a basic rough cut I just consolidate/transcode to dnxHD and then after I got mine final sequence I EDL to an grading program (Davinci, After effects). Doing a rough cut and then transcoding saves you a lot of time and disk space. But this is assuming that your workstation and NLE can smoothly handle the AVCHD material.
 
I dont see how transcoding 4:2:0 to 4:2:2 is going to help.

The extra data is only padding the bits.

I have CC and graded plenty of broadcast work from AVCHD (4:2:0) in Resolve - and it works fine.

Have also done the same with dual recorded footage from a Samurai (4:2:2) and yes its better but I dont see the transcode is going to give that advantge.
 
Any "new" data that is going to be created while grading is going to be created in a better resolution. The 10bit nature of the codec is also going to help when grading. And for me there is a big difference when grading in After effects if I use a i-frame or a compressed long gop codec.

Resolve works in 32bits so transcoding to a 10bit codec isn't going to help bit-wise as the software already proceses the footage in 32bits...
 
Any "new" data that is going to be created while grading is going to be created in a better resolution.
Click to expand...

So the "new data" is created from what color information in the 4:2:0 source file when transcoding to 4:2:2.
 
Yes, it's derived from 4:2:0 but that doesn't or shouldn't stop the program from creating new material in 4:2:2. I don't know, I'll make a test and post some stills here.
In any case it isn't just the colour resolution the compression and the bitrate should also play a role. Maybe not so much in just a single push but when you stack more and more effects (corrections) one uppon the other it should matter.
 
Last edited:
Oh now I see what you meant! No I wasn't talking about new data being created in the process of transcoding but in the process of grading.
 
It was my understanding that, while it wouldn't make anything "better", it WOULD make the image stay together (or not fall apart as badly) during heavy grading. (as compared to the standard AVCHD codec) Have I got that wrong?
(and if I do, what would the point be for the conversion tool in the first place if it provides no benefit?)


ps Merry Christmas!
 
yes, think like if you are doing math with only one decimal instead 2 or 3.

Some operation can be ok, while some could be destructive.
So if you divide 1 by three it gives .3 or 0.333 , that is not a big deal if you stop here, but if you want to multiply back by 3, then you get 0.9 or 0.999.
it is not a big deal either, but if you apply usual rounding method, 0.9 stay 0.9 and 0.999 become 1 at worse.
do this several time and you get significant difference at the end.
 
Right, that's a nice analogy.

@Chris: do a test! Grade the FS100 footage. Grade the same shot in AVCHD and in Cineform. Apply several different grades one upon the other and then compare the results. Then do a second test: export the graded footage (export the AVCHD as AVCHD and Cineform as Cineform) and then bring it back in and do some more grading. Compare the results.

There really is no point in asking if you can do a test yourself and see what quality standards which workflow ticks for you. These are your standards so we can't make a decision for you:)
 
Last edited:
That video is a great find. I won't re-hash what's already been said in this thread, but I will recommend Steve Hullfish's book The Art and Technique of Digital Color Correction. If you really want to get serious about color correction, you should definately check it out.

Also, I concur with those that have said it's easiest to color correct so all your shots match and are consistent, then apply the grade.

-Mike
 
KINOKS said:
Right, that's a nice analogy.

@Chris: do a test! Grade the FS100 footage. Grade the same shot in AVCHD and in Cineform. Apply several different grades one upon the other and then compare the results. Then do a second test: export the graded footage (export the AVCHD as AVCHD and Cineform as Cineform) and then bring it back in and do some more grading. Compare the results.

There really is no point in asking if you can do a test yourself and see what quality standards which workflow ticks for you. These are your standards so we can't make a decision for you:)
Click to expand...

Thanks, I was looking for good book on Color Correction and I just bought that one. I have shot for a long time, but I really have not gotten really hands on with doing much CC and grading. I am on a crash course now to get more up to speed on it. Thanks again.

I also shoot with an FS100. I have been using Frank's PP G-Log A. I do not light flat, but what I have found is that this more neutral PP gives me advantages in challenging lighting situations much like the old film technique of flashing did. Because protecting your highlights is so critical in digital video, using the more neutral PP seems to me to hold the shadow detail together pretty well so that I can expose for the highlights and I know that I am not losing everything in my shadows. I started by shooting reversal film, which also required exposing for the highlights, but the baked in contrast of reversal film (It was designed for direct projection) always bugged me ( I hate crushed blacks). Now with the FS100, I can expose the same way and I am getting my contrast where I want. I have been doing the limited amount of CC and grading just in FCP, and while I have been pleased with the results, I really want to work towards more finesse.
 
You must log in or register to reply here.
Back
Top