Cancel that thought...

[Tim van Dammen]

Apparently according to this thread I don't know what I'm doing.

So I'd love to have someone tell me the settings to get my Red to look like Alexa!

Andrea and Chris crack me up - its like a religious debate or political debate... circles meaning nothing.

 

[Jason Sinclair]

This would be a great script. What was your childhood like? -only joking. Thanks for your honesty.

 

[Marc Wielage]

I'm not only talking about me, I'm referring to 2 different colorists who tell me how much extra work it takes for them to get a decent look from EPIC because of all the secondary work needed to get the colors balanced, most notably yellows and skin tones.

Putting on my colorist hat, I haven't found this is a problem with either Baselight or DaVinci Resolve. I might make a mild custom curve adjustment and tweak the secondaries and overall saturation a little bit, but the rest of it looks fine to me. I haven't seen that much of a skin tone issue to really worry about it that much, and it's not that much extra work at all, once you drop in a preset.

To me, this is no different than setting up a "base correction" for film transfers. It's always been this way in color correction.

 

[Bruce Allen]

Putting on my colorist hat, I haven't found this is a problem with either Baselight or DaVinci Resolve. I might make a mild custom curve adjustment and tweak the secondaries and overall saturation a little bit, but the rest of it looks fine to me. I haven't seen that much of a skin tone issue to really worry about it that much, and it's not that much extra work at all, once you drop in a preset.

To me, this is no different than setting up a "base correction" for film transfers. It's always been this way in color correction.

Mark, if you want to do a tutorial or sell me that Resolve preset / curve, I'm game

For example, the skin tones in the latest Spiderman trailer have that "RED look" to me anytime it's not controlled daylight light sources:
http://www.reduser.net/forum/showthread.php?78564-The-Amazing-Spider-Man-Trailer-3

The very first closeup of Peter Parker, for example:
- entire face seems one shade of brown, except for greyish side highlights and vaguely reddish ears and cheeks
- except for lips which look like the wrong color pink
- whites of eyes look oddly pink and dull
- and hair kinda blends into face in unappealing mushy way

I mean... how the heck do you correct for that?

Sorry, should probably take it to the other thread. But seriously, would love to hear what you as a colorist use as a starting point for RED footage.

Personally for me it's:
- balance out as best I can
- increase saturation a lot overall
- increase saturation some more in certain areas (sometimes it's the shadows, sometimes it's in the relatively-desaturated areas, etc)
- do secondary correction on reds and greens to make them more pure and less pink / brick color (RedColor3 has helped with this)
- by now the hair has usually become a kind of blue shade. So I go back in and fight to get the hair looking somewhat like human hair
- Stop, take a breather and look at some pictures of real faces shot on real film.
- push on...

Any suggestions for improvements in workflow would be appreciated!

Thanks, man.

Bruce Allen
www.boacinema.com

 

[Marie Turock]

YES!! Posts at 3:00AM rule! Just made one. Can't say I know of any other company that has such an online omnipresence. Also, don't know of any other camera that can do what Scarlet does for me at the price it sells for. Thanks for doing it right.

 

[M Most]

For example, the skin tones in the latest Spiderman trailer have that "RED look" to me anytime it's not controlled daylight light sources:

You're judging skin tones in a huge tentpole feature based on a heavily compressed clip on the Internet, viewed on a computer monitor? And probably an uncalibrated one, at that??

I would strongly suggest waiting until the picture is released, viewing it on a large screen in a theater like it's meant to be viewed, and forming your opinion based on that. That at least gives the production and post production team a fighting chance to do its job and show you the results sensibly.

 

[M Most]

Any suggestions for improvements in workflow would be appreciated!

Here's a few:

1. Keep it simple. Don't try to correct for specific things until everything that's basic is as perfect as it can be.
2. Stay away from lift, gamma, and gain if you're using a log original (i.e., RedlogFilm). Stick to exposure (aka offset) and contrast correction, both done prior to a normalizing LUT. Balance using the overall "printer light" approach, correcting almost exclusively for flesh tones if there are some in the image. Everything else will largely fall into place. Use a bit of shadow correction to rebalance the blacks if you want perfectly balanced blacks (although you don't always want that......).
3. Use primary corrections through tracked windows rather than keyed secondary corrections whenever possible. You'll avoid noise and other very nasty artifacts.
4. Don't assume that everything in the image should look a certain way unless you have evidence that it should. If you use the simplified approach I've outlined above, the colors captured by the camera should make themselves very obvious. By avoiding use of lift gamma and gain for the basic correction, you're no longer "pulling the picture apart" and contaminating certain colors in the process. The colors retain their original values and the pallette is noticeably wider. In other words, your base correction is much less destructive.
5. Do any further tweaking of the original correction downstream of the normalizing LUT. This gives you good control and better separation for any isolations you might want to do at that point.
6. Accept the fact that all digital cameras have somewhat different inherent visual characteristics, and all of them are different than film. Embrace their particular aesthetic and don't try to make them look like something they're not.

That's what I do, anyway...

 

[Bill Anderson]

How about a RED tutorial series: Post with Most. I'd buy that. ( :

 

[illya laney]

Well, Canon just announced yesterday that they will be showing their 4K C500 camera with a workflow presentation, a 4K short shown in 4K, and some hands-on touchy feely with the camera and accessories one week from now in LA at the LA Film School if anyone wants to go and compare whatever they are offering with the RED experience. I'm not saying anyone should go or needs to go, but given the ongoing discussion, I would assume some of you would like a comparison of whatever they are presenting vs whatever you yourself have experienced with your use of the RED cameras.

I believe you're talking about the short film they showed at NAB called "Man and Beast." It was shot by Cronenweth and Fotokem developed the workflow.

 

[Gavin Greenwalt]

2. Stay away from lift, gamma, and gain if you're using a log original (i.e., RedlogFilm). Stick to exposure (aka offset) and contrast correction, both done prior to a normalizing LUT. Balance using the overall "printer light" approach, correcting almost exclusively for flesh tones if there are some in the image. Everything else will largely fall into place. Use a bit of shadow correction to rebalance the blacks if you want perfectly balanced blacks (although you don't always want that......).

Ummm, or you could just do a 32bit conversion to linear perform those operations and return to log space when you're done. Whether or not your image is log-source or lin-source is irellevant seeing as we can easily switch which space you're performing the correction in and back.

 

[Elsie N]

Here's a few:

1. Keep it simple. Don't try to correct for specific things until everything that's basic is as perfect as it can be.
2. Stay away from lift, gamma, and gain if you're using a log original (i.e., RedlogFilm). Stick to exposure (aka offset) and contrast correction, both done prior to a normalizing LUT. Balance using the overall "printer light" approach, correcting almost exclusively for flesh tones if there are some in the image. Everything else will largely fall into place. Use a bit of shadow correction to rebalance the blacks if you want perfectly balanced blacks (although you don't always want that......).
3. Use primary corrections through tracked windows rather than keyed secondary corrections whenever possible. You'll avoid noise and other very nasty artifacts.
4. Don't assume that everything in the image should look a certain way unless you have evidence that it should. If you use the simplified approach I've outlined above, the colors captured by the camera should make themselves very obvious. By avoiding use of lift gamma and gain for the basic correction, you're no longer "pulling the picture apart" and contaminating certain colors in the process. The colors retain their original values and the pallette is noticeably wider. In other words, your base correction is much less destructive.
5. Do any further tweaking of the original correction downstream of the normalizing LUT. This gives you good control and better separation for any isolations you might want to do at that point.
6. Accept the fact that all digital cameras have somewhat different inherent visual characteristics, and all of them are different than film. Embrace their particular aesthetic and don't try to make them look like something they're not.

That's what I do, anyway...

Copied and saved.

 

[M Most]

Ummm, or you could just do a 32bit conversion to linear perform those operations and return to log space when you're done. Whether or not your image is log-source or lin-source is irellevant seeing as we can easily switch which space you're performing the correction in and back.

What you're talking about might apply if you're using a compositing program, where you have to explicitly do those transforms if you want them used. It doesn't apply if you're using a typical color grading program because you don't control how the internal pipeline works, although in most cases it works in linear light and very deep bit depth (32 bit float is somewhat standard). In a typical color grading program, you are presented with a set of "video" style tools (i.e., lift, gamma, gain), a set of "log" type grading tools (i.e., exposure, contrast, and shadows, midtones, and highlights, usually with individual pivots for each), or both. You choose one or the other or a combination of the two based on the material you're working with and your own particular grading style.

Your statement is not really relevant when using typical grading software such as Resolve, Baselight, Lustre, Scratch, Pablo, or any other "standard" grading toolset, which is what I was referring to and what the post I was responding to was asking about.

 

[Philipe Ratton]

Copied and saved.

As usual

 

[Elsie N]

As usual

Heh, heh... I've got Mike's, David's, and others' instructional quotes copied and saved on at least three different computers. Now all I have to do is assemble them in one spot for easy reference.

Good to see I'm not the only one paying attention. '-)

 

[Zachary Marsh]

No.. In the end "people" paid over $1 billion to see Avatar. If by "people' you mean your friends, I can't comment, but "people" like and will pay for good imagery, and stay away from the dreary and the ugly in droves.

As to Paranormal Activity the particular story justified the ugly visuals. Few stories have that angle, and you would not want only movies with that kind of angle to come out. A romantic comedy looking like that would be unwatchable to most.

As for propaganda, Zacuto makes major $$ selling DSLR rigs to wedding type shooters. They are not about to tell their customers that their favorite toy is mediocre.

You're hilarious. How many RED films should I list that were critically and financially bombs?

Che didn't even make back it's budget. Knowing is critically considered stupid (34% Rotten Tomatoes). Gamer didn't make it's budget back, and has a 30% on Rotten Tomatoes. The Girlfriend Experience, Valhalla Rising...It didn't matter what camera was used in these cases. District 9 was shot on RED and did well. Why? THE STORY.

Avatar made money because it's emotionally manipulative and the James Cameron is a master of that. The story wasn't original or good, people were more interested in the world. It's an exception, not a rule.

Notice how many successful and unsuccessful movies are shot on a variety of cameras? As long as people can see what's happening on the screen and are engaged, that's all that matters. Yes, there's a lot that goes into that, and yes having a great image can contribute to that, but it is hardly a requirement. Which you've admitted to with Paranormal Activity's success.

 

[Joseph Hutson]

Here's a thought. Forget about tests and go out and shoot something on an iPhone or whatever camera you can get your hands on. Would that not be more profitable than talking about companies with successful business plans that you don't like?

At some point, I think it's not about what camera you are shooting with. It becomes just shooting.

The camera nobody talks about in tests is the camera that played a big part in me getting my current job. I'm not saying the iPhone is what you want to shoot a big movie on, but it is a camera.

 

[Michael Millichamp]

Here's a thought. Forget about tests and go out and shoot something on an iPhone or whatever camera you can get your hands on. Would that not be more profitable than talking about companies with successful business plans that you don't like?

At some point, I think it's not about what camera you are shooting with. It becomes just shooting.

Can someone put this man in the favorite reduser quotes?

 

[Bruce Allen]

Here's a few:

1. Keep it simple. Don't try to correct for specific things until everything that's basic is as perfect as it can be.
2. Stay away from lift, gamma, and gain if you're using a log original (i.e., RedlogFilm). Stick to exposure (aka offset) and contrast correction, both done prior to a normalizing LUT. Balance using the overall "printer light" approach, correcting almost exclusively for flesh tones if there are some in the image. Everything else will largely fall into place. Use a bit of shadow correction to rebalance the blacks if you want perfectly balanced blacks (although you don't always want that......).
3. Use primary corrections through tracked windows rather than keyed secondary corrections whenever possible. You'll avoid noise and other very nasty artifacts.
4. Don't assume that everything in the image should look a certain way unless you have evidence that it should. If you use the simplified approach I've outlined above, the colors captured by the camera should make themselves very obvious. By avoiding use of lift gamma and gain for the basic correction, you're no longer "pulling the picture apart" and contaminating certain colors in the process. The colors retain their original values and the pallette is noticeably wider. In other words, your base correction is much less destructive.
5. Do any further tweaking of the original correction downstream of the normalizing LUT. This gives you good control and better separation for any isolations you might want to do at that point.
6. Accept the fact that all digital cameras have somewhat different inherent visual characteristics, and all of them are different than film. Embrace their particular aesthetic and don't try to make them look like something they're not.

That's what I do, anyway...

THANK YOU! Ah man - I have a box of Sam Adams sitting at my apartment that I was supposed to give to you on behalf of Gunleik and myself as thanks for your help with Resolve and R3D consolidation that I haven't had time to drop off.

Now I owe you more thanks. Great. I should have just made it a crate or built a small structure out of beer kegs and delivered that instead. OK damn it. Need to get more. You prefer regular Sam Adams or their seasonal stuff?

You're judging skin tones in a huge tentpole feature based on a heavily compressed clip on the Internet, viewed on a computer monitor? And probably an uncalibrated one, at that??

I would strongly suggest waiting until the picture is released, viewing it on a large screen in a theater like it's meant to be viewed, and forming your opinion based on that. That at least gives the production and post production team a fighting chance to do its job and show you the results sensibly.

I think that people's computers are a key viewing environment for trailers.

Maybe I'm biased because our last trailer for The Avengers is the new record holder for most internet downloads in a day

Personally, I watch trailers at work on my side computer - a 2011 24" iMac with stock monitor settings - eg emulating a good consumer computer viewing environment. When at home, I watch trailers on my Sony projector, my calibrated DreamColor monitor and / or 27" iMac.

I just wish - speaking as a RED nut - that I couldn't spot every movie shot on RED by its "RED colors!" look...

For contrast, I remember saying "this likely wasn't shot on RED" 5 shots into the trailer of Hunger Games. Which I watched at 360p on Youtube. The rich, natural colors instantly read as filmic, even at low res: http://www.youtube.com/watch?feature=player_embedded&v=mfmrPu43DF8

I'm sure it doesn't bug many folks of course except for crazy old me - Spiderman trailer looks fantastic otherwise and marketing and filmmakers are doing a great job - seriously, wow. I'm there opening weekend.

Bruce Allen
www.boacinema.com

 

[Marc Wielage]

1. Keep it simple. Don't try to correct for specific things until everything that's basic is as perfect as it can be.
2. Stay away from lift, gamma, and gain if you're using a log original (i.e., RedlogFilm). Stick to exposure (aka offset) and contrast correction, both done prior to a normalizing LUT. Balance using the overall "printer light" approach, correcting almost exclusively for flesh tones if there are some in the image. Everything else will largely fall into place. Use a bit of shadow correction to rebalance the blacks if you want perfectly balanced blacks (although you don't always want that......).
3. Use primary corrections through tracked windows rather than keyed secondary corrections whenever possible. You'll avoid noise and other very nasty artifacts.
4. Don't assume that everything in the image should look a certain way unless you have evidence that it should. If you use the simplified approach I've outlined above, the colors captured by the camera should make themselves very obvious. By avoiding use of lift gamma and gain for the basic correction, you're no longer "pulling the picture apart" and contaminating certain colors in the process. The colors retain their original values and the pallette is noticeably wider. In other words, your base correction is much less destructive.
5. Do any further tweaking of the original correction downstream of the normalizing LUT. This gives you good control and better separation for any isolations you might want to do at that point.
6. Accept the fact that all digital cameras have somewhat different inherent visual characteristics, and all of them are different than film. Embrace their particular aesthetic and don't try to make them look like something they're not.

Good job, Mike, and well-said. Red should make this part of the manual!

I must know at least 50 colorists in LA, and to me, the good ones will generally agree that very often, "less is more." Start with a reasonable correction, and then when you really need to go wild, make those the later layers / nodes / whatever.

This was my reason for several conversations in the Resolve area, where people are trying to do 9 nodes for a relatively simple shot of two people standing by a tree. I've done shots with 20 layers or windows on occasion, but those where whacked-out, crazy, screwed-up situations where we were basically trying to save a very problematic scene or introduce a very heavy, complex look. A straightforward scene, shot well, shouldn't take more than 1 minute and a couple of nodes to make it look reasonable. Nothing's worse for a colorist than when you go through a session and realize that node 7 basically undoes what was done in node 2, or node 8 just adds noise to compensate for a bad balance in node 1. Color correction can really be a house of cards if you aren't careful.

I honestly didn't think the new Spiderman trailer looked bad at all, and there's always the chance that blotchy, uneven skintones are the fault of bad makeup. In a movie I worked on some years back, Queen of the Damned for Warner Bros., we spent a lot of the color-correction process just fixing makeup problems (on the two lead actors), at least 100 hours mainly solving that problem. You can pull keys and even out blotchy skin, but at some point, it becomes more of a visual effect than color-correction, per se.

I'm always amazed by what certain people zero in on during a shot. The director will worry about a glass window in the back; the DP is worrying about the second actor being slightly out of focus; I'm trying to look at the entire shot and how it's dark on one side. None of us are wrong. These are very subjective areas, and our brains are sometimes larger factors than cameras or monitors.

Back to the original topic: I would be very curious to see what would happen if you took 7 different high-end digital cameras, shot them on the same set with a decent DP, then went into one color correction room and spent a week making all the cameras look exactly the same, as much as humanly possible. At least then, the other characteristics would be on an even playing field, and nothing would jump out and look awful or weird.

I've asked the Zacuto people before if they were interested in releasing their past tests for sale on Blu-ray, and so far, they haven't been interested. Me, I'm skeptical of what I can really see as streaming media on questionable computer displays (even my own). A Blu-ray on a decent consumer monitor would be a lot better than that; a 4K theatrical display would be even better, from a traditional DCP.

 

[geoffboyle]

I hope that everyone can do some testing on their own that includes all aspects of image capture. We just aren't up for the grief. We have better things to do.

There are lots of options for you to consider. I hope that RED is one of them.

Jim

Welcome to my world Jim

It's not that easy!

I'll be doing a couple of comparisons in the next few months, one in Seoul and one in Hannover.

At the moment it's Epic, Alexa & F65 although we'll see what else I can get hold of!

As usual with my tests we will concentrate on dynamic range and colour response.

I'll take advice from everybody as to how to finish them as I always do.

There will be many witnesses/participants to these tests.


OF COURSE i'LL STILL GET IT WRONG AND GET ABUSE FROM EVERYBODY