Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Resolve Tutorials

Warren Eagles said:
Just uploaded Grading Arri Alexa PT 2
http://vimeo.com/channels/icolorist

Warren
@warreneagles
Click to expand...


Warren, while I certainly appreciate the effort you're putting into educating up and coming colorists, I have to say that as an experienced colorist, I really disagree with the approach you're advocating on this particular clip.

A log image is not an image, it is a container. It is designed simply to carry more relevant image data than a 10 bit image normally can by applying a non-linear gamma curve of a specific design that maximizes levels available to useful detail and minimizes the levels used by less noticeable detail. It is not designed to be graded by a video style color corrector directly because the values in the image you're starting with are not "normalized." They are skewed toward what I just described. If you use the grading approach you're advocating here, you will likely have a result in which the upper midrange is too high (leading to "flat" faces, for one thing) because you haven't yet normalized it. The only real way around that is to crush the blacks rather aggressively in order to create contrast in the upper midrange of the picture. Crushing the gamma is not a good way to do this because the gamma control is going to affect the middle of the overall image curve much more than the extremes. That's fine in a video gamma image, but it's not fine in a log image because the information you want to be affecting is not necessarily in the middle of the curve.

You demonstrated an approach that basically says "LUT or no LUT," but that's not the way most of the colorists I know - including myself - see it. A LUT essentially represents a curve that "normalizes" the values in the log container to display properly - in terms of grey scale - on a video or projection device. At that point, you can use video scaled controls to properly grade the image. But under most circumstances, LUTs are used in a DI style pipeline - in other words, they are put at the output of the base grade, not before. Putting them before defeats the entire purpose of recording in a log format in the first place. In most cases, what we do is apply a grading node, then the LUT, then additional grading nodes as needed. That way, if the LUT is a bit aggressive, we correct the image prior to it hitting the LUT so that what is coming through it is correct. At that point, we have a "normalized" video image that reacts properly to video grading controls and yields good isolations as well. In addition, in the case of Arri Alexa footage, it also implements a color matrix that yields proper saturation and more accurate colors, something you would have to simulate using extreme amounts of saturation in the grading system and some secondary manipulations if you don't incorporate it. Those techniques add noise and cause unnatural results. You can, of course, use a curve rather than a LUT for the normalization, but you won't get the benefit of the matrix (unless you're using Red footage, of course, where it's incorporated in the debayer process).

I know a number of colorists - both experienced and less experienced - who use the approach that you're advocating here. But that approach is based on what we did with video images for all of these years. That's just not appropriate or effective in a situation in which we're not dealing with video images anymore. Log formats are best used in a digital intermediate style pipeline, in which they're treated as the information containers they actually are, rather than as video images that they're not. That takes a bit of understanding as to what's being dealt with, and what a LUT (or a curve) is bringing to the table. I know you understand all of this, but you might want to think about presenting that alternative in future videos.
 
M Most said:
…under most circumstances, LUTs are used in a DI style pipeline - in other words, they are put at the output of the base grade, not before. Putting them before defeats the entire purpose of recording in a log format in the first place.
Click to expand...

(bold mine not Mike's)

This point cannot be emphasised enough, and is one that so many people seem to miss.
 
M Most said:
But under most circumstances, LUTs are used in a DI style pipeline - in other words, they are put at the output of the base grade, not before. Putting them before defeats the entire purpose of recording in a log format in the first place.
Click to expand...
Are you coming at this from a Resolve point of view or a Baselight point of view?

I know with Baselight, what we generally did was load in the source files, select the appropriate Truelight LUT for whatever color space we need, add a base correction layer, then do the final color-correction on an additional layer(s). Are you advocating using a LUT as a layer or node prior to all color correction, or as an overall preliminary set-up?

I'm curious how something like Truelight would fit into a Resolve workflow -- or if it would have to be done in a completely different way. All the Resolve stuff I've done so far has been Rec709 in & out. I haven't done a workflow test to see how input LUTs and output LUTs would apply in that situation, nor have I tried Cinespace LUTs yet. The section in the Resolve manual starting on page 72 goes into it in a superficial way, but there's a lot of steps left out from my perspective.

I think I see both your point and Warren's point. The question is whether you could engage all the parameters in a correction to precisely emulate a 3D LUT, and to me, it'd be difficult to get there 100%, so I agree with your general statement. But a simple LOG -> LIN LUT would get more than half the way there, just in terms of giving you more range and not having to slam the corrections (or build multiple nodes) to get an acceptable picture. In Warren's case, if he was only delivering Rec709 material, his method would work, but to me it wouldn't be ideal.
 
i bought these lessons for resolve and they are great for DIT's that want to dig deeper into the colorist world as we are constantly being tasked to do more and more of this work on set, in real time. i do not claim to be a colorist and probably never will or want to be, but i like to make the images that come off my cameras look nice for me and for the client, with no illusion that what i am doing on set is even close to the final, but at least it is a step in that direction, and in 99% of the time it makes everyone on set very happy to see pretty picture that are not just on the camera monitor, but out in the world on other devices.

highly recommend watching these classes.

dino
 
I did the Resolve training with Warren in Singapore at Media Village. Highly recommended as well. Warren is a great guy and he knows what he is talking about.

M Most: That sounds interesting and seems to make sense. However, without a proper example why applying an output LUT is better it remains a bit theoretical. My experience in Resolve is that sometimes things just work fine when in theory they shouldn't. Any chance to show us what you are talking about? 'Flat faces' vs 'proper faces'?
 
Try it yourself with any LogC material you happen to have. The difference is obvious.

You might want to make a version of the Arri LUT that has a bit less green/yellow in the gain. You only need a small correction to make for a more accurate LUT for most photography. That's what we did, and have found that the resulting modified LUT works almost perfectly regardless of the show.
 
Marc Wielage said:
Are you coming at this from a Resolve point of view or a Baselight point of view?

I know with Baselight, what we generally did was load in the source files, select the appropriate Truelight LUT for whatever color space we need, add a base correction layer, then do the final color-correction on an additional layer(s). Are you advocating using a LUT as a layer or node prior to all color correction, or as an overall preliminary set-up?
Click to expand...

I'm coming at it from a proper image processing point of view. It doesn't matter what grading system is involved. They all allow you to place a LUT before or after a grade, and in some cases - Resolve and Baselight included - allow you to place a LUT anywhere in the processing path.

If I'm reading what you're saying correctly, I don't know why you would do what you say you did. In Baselight, I would set it up exactly the way I described: the load strip, followed by a base grade layer (probably using a Film grade at that point), followed by the LUT, followed by anything else (usually a video grade at that point). Using the LUT at the beginning of the chain puts you in the same situation as tape to tape color correction. Do you really want that when you can have so much more flexibility?

I think I see both your point and Warren's point. The question is whether you could engage all the parameters in a correction to precisely emulate a 3D LUT, and to me, it'd be difficult to get there 100%, so I agree with your general statement. But a simple LOG -> LIN LUT would get more than half the way there, just in terms of giving you more range and not having to slam the corrections (or build multiple nodes) to get an acceptable picture. In Warren's case, if he was only delivering Rec709 material, his method would work, but to me it wouldn't be ideal.
Click to expand...

It wouldn't work particularly well if he is using Alexa material without a matrix applied, which is what he seems to be doing. That would not only force more extreme corrections, it would likely also force a rather significant saturation boost. As I said, Red footage coded to RedlogFilm wouldn't have that problem.
 
Christopher Barrett said:
Heh... after watching the Resolve tutorials... I now read Warren's posts with an Aussie accent.
Click to expand...

Actually, while Warren has lived in Australia for +10 years he is from the UK, and most Aussies I know would say he has a London accent.
Either way the more guys and gals who offer their advice on grading, and using Resolve the better.

Peter
 
Hey Warren, I loved the Resolve course, fantastic stuff!

You sometimes refer to Center-Click, which I haven't figured out how to do. I keep my mouse set to just left/right click and can't figure out what you are doing to Center Click.

Silly problem to have but there I am.

Thanks!

Jim Hare
Sydney
 
jimhare said:
Hey Warren, I loved the Resolve course, fantastic stuff!

You sometimes refer to Center-Click, which I haven't figured out how to do. I keep my mouse set to just left/right click and can't figure out what you are doing to Center Click.

Silly problem to have but there I am.
Click to expand...

You really need a three button mouse to run Resolve effectively if you're running it on a Mac.
 
Good on yer, mate ...

Good on yer, mate ...

Peter Chamberlain said:
Actually, while Warren has lived in Australia for +10 years he is from the UK, and most Aussies I know would say he has a London accent.
Either way the more guys and gals who offer their advice on grading, and using Resolve the better.

Peter
Click to expand...

Agree wholeheartedly, Peter ..... I've been following Warrens's Resolve 8 tutorials on fxphd and they're damn good, no matter what accent he uses.

As an ex-Brit who lived in Sydney for a couple of years I too quickly sussed out that Warren was some kind of Wanstead bovver boy who learned to speak 'strine ... accents apart, if you're interested in understanding the power of DaVinci Resolve 8 you should definitely check out Warren's tutorials.

Keep up the good work, matey .... and btw, Peter our three colorists are really enjoying the functionality and elegance of Resolve 8.1 ... thanks for getting us on the beta program.

Cheers,
Neil
 
M Most said:
You really need a three button mouse to run Resolve effectively if you're running it on a Mac.
Click to expand...

So no command-key combination to simulate a center click? Either have it or you don't?

I would have thought center click would just be a shortcut for something else, like option-click, command-nose-on-tab-key etc.

I'm just looking to apply the tips from Warren's FXPHD course.
 
I bought Warren's tutorials from FXPHD and as a colourist of 15 years or more of experience they really taught me the basics of resolve and what the buttons did. This course is the best $200 I have spent in a long time.

The course is great for beginners or experienced people like myself so if you buy Resolve, buy the course also.

And Warren didn't pay me to say this.

For the record my first proper job on Resolve has been colouring a feature film that I have almost finished, and I had 7 days to finish (1:30 feature).
His course brought me up to speed in no time and the director is happy as a pig in shit!

And.... I have only had my box/ resolve for 2 weeks.. So if that isn't a good rap about Warren's product and resolve I don't know what is!
 
jimhare said:
So no command-key combination to simulate a center click? Either have it or you don't?

I would have thought center click would just be a shortcut for something else, like option-click, command-nose-on-tab-key etc.

I'm just looking to apply the tips from Warren's FXPHD course.
Click to expand...

A three button mouse can be had for probably less than $10, which is a pretty paltry investment even if you only use it for one program.
 
The centre click is on the apple mouse with the little ball. You press the little ball down to centre click, I don't think you can buy those mice anymore but as Mike is saying you can buy any 3 button mouse from a place like officeworks for not much.
 
You must log in or register to reply here.
Back
Top