cs5 color and workspace help

G

greymog

Well-known member
Hey everyone, I didn't know where exactly to put this thread, I saw color here so I thought it would be the best place, mods pls move if I'm wrong of course.

Here's my thing. I read a lot of prolost and that's basically what got me into this in the first place years ago. I started reading it again for good old times sake, and with time more things are making sense, but my issue is with lin and log.

I've learned the difference between linear light values and logarithmic ones, how our eyes see light, similarly how film sees light, and about the linear nature of how it behaves as opposed to how i perceive its behavior. My problem is implementing this in a workflow for coloring and shooting. I don't know if I would use it with the cameras I have access to now, but still I want to know what's going on in AE CS5 color management and why I pick spaces that I pick or not etc.

Stu had enclosed a white paper for cs4 color management, and I read it, and there are basically workflow steps in there that I'm interested in, particularly the digital cinema one, but there's so little I know about why I may or may not take these steps.

First of all, I'll try and give the impression I have, mistakes and all, this is what I'm thinking so far. I select the workspace in order for my processes to respond within a range suitable/assigned to a certain output. Let's say I pick universal camera print density as a working space so that I can accommodate for the option to print to film, export to a dcdm file (which I also have questions about), and take out 'dailies' for dvd or hd viewing on a conventional monitor.

So that's workspace. But I still don't know why I choose one space over the other except for 'that its the suitable one for the output'. What's happening to values? This is where my confusion kicks in, when we start interpreting footage. My situation is always an 8-bit file, either rec709 or rgb. I have shot red, and things are making more sense reading this, still I also (more frequently these days) have r3d's to deal with along with those dslr/ex1/3 files.

Do I change their profile to match the working space's profile? Or do I assign them the embedded profile? Is the profile telling the workspace 'Hey workspace, I'm an srgb file, interpret me to your numbers as such', or 'hey workspace, i don't feel good being an srgb, change me to match you so I respond appropriately'. I'm confused as to what to assign to my footage, and if I'm working with a red file, I'm confused as what to put it out as after demosaic and why.

Also, when it comes to output module, I'm confused as to what happens when I assign. If I'm correcting for a film print, and I want to export an hd version for a rec709 monitor, should it look like my compositing screen when I open it somewhere else? so confused.

Why am I linearizing my gamma in ae also? I mean, all I do is grade, very, very, very little green screen work, no 3d work, almost no compositing. I just cut in an NLE and want to finish to a good grade that can hold up. I never care what camera I use, but I'm always limiting my actual scene's dynamic range, I'll gel my windows all of em, I'll fill every shadow till my histogram doesn't clip white or black. That's my goal for capture.

Linear light hits the sensor, that's how it's mapped, but we don't see it like that I get it. Most cameras on my list except the red are baking in a look right away, so I'm never getting the linear representation of those values right? They being chucked when packed into a curve that allows simulates my perception (logarithmic right?). So why am I doing this with anything less than a cineon log file or an slog file or a red log or panalog file?

I'm so effin confused about this. Here's what I need as a shooter...

I need to shoot and make sure that my stuff doesn't clip ever on either side of the histogram, and (PLEASE correct me if I'm wrong) I believe that footage can handle a lot more outputs maybe even a print if treated like this. I need to know what to do in cs5, but more importantly why, in order to accommodate a digital cinema workflow and have the option to export to either a rec709 output or a dcdm output. Clients are asking me for these files recently, and I'm stuck.

I know I've been doing it wrong before, I wanna do it right. This is a huge post I'm sorry but there's so much in my head this week I've been drowning myself in this stuff. Someone help clarify all this. Thanks to all for their time.
 
Another question, maybe clearer...

I have footage coming in with an embedded profile of rec709. I still don't know WHY I'm in universal film print space, instead of rec709. Why would I choose univcamfilmprintdensity, except for accommodating for a film out? Also, how do I accommodate for a rec709 output using this workspace? Just the output module looks different in result, I'm confused.

Also, I'm working in a linear space. I'm hitting linearize in project settings. I have footage coming in, that same rec709 footage and I know it's not linear. Or is it? I keep it in its embedded profile, which is different than my workspace? Also, do I use the converter to convert it from log to lin? I don't know why I'm linearizing, again I just grade and correct and apply look effects I barely green screen and barely composite.

Sorry that this is so long I am obsessing about this.
 
Can one think of making dcdm exports if one shoots an 8-bit 4:2:0 long gop capture if one doesn't clip on either side? Or is it just futile and why?
 
Here's the deal. Your file is just a bunch of numbers. True, these numbers are in a specific format. All that means is that a piece of software made to read that format can read your file -- that is, the numbers are where the software expects to find them. But it's still just numbers.

To turn those numbers into visual information, something (hardware, software, or a combination) has to translate those numbers into colors, and hand that off to a display system to show those colors to you.

Now the fun part -- how do you make sure that the numbers in your file always display as the colors you want? For example, how do you make it so that the same fire-engine red displays on an HDTV screen, your NLE's computer screen, and film?

That task is what device profiles are for. The profiles are used to translate your file's numbers into the numbers needed to make the display device display the colors you want. Hardly anything will display your file "raw" without translation, because every device, from your on-camera electronic viewfinder on, uses different phosphors, and different ways of driving those phosphors, so that the same number gives wildly different colors from device to device. So just about *every* device has a profile or working space to translate your file's numbers into the colors you expect to see.

This is why you have to choose the profile you want to use, and why you have to know what kind of device you are using to display your file.

What makes this confusing to many is the use of jargon. In particular, the overlapping of the concepts of a working space with a profile. They are not the same, but you can think of them as being different names for the same concept. If you just think of them (profiles, working spaces, whatever) as being translators, you should get the general idea.

If you want to go deeper than this, start looking for a book on color theory. It's a big complex subject. Getting the colors right on every output device isn't easy. But even as complex as it is, it's now fairly well understood, and the software is making it easier with every new software generation.
 
Thanks Bruce... really. I've been up for weeks. I do want to get into this, I love figuring stuff out and I may be asked to do this quite intensely in the near future for a living. I need to get some things straight in my head.

My footage is almost always coming from either 1) some dslr, either an mts or an h264 (you can guess which cams these are) and ex1/3, a RED one, and maybe, just maybe once in a blue moon, an alexa.

I use aftereffects with colorista and the native utility controls if need be, very simple stuff after edit. Someone I know takes care of compositing almost exclusively for me and he's gonna chime in on this thread this week, his name is nash.

I still want to know why I linearize my workspace. I know my footage is coming is not as a linear record, but mostly in vid space (this is for the ex's and dslrs) but I may be wrong. They have srgb and rec709 profiles embedded. I'm assuming this is vid space.

If I make my project space linear, say universal film print, or rec 709, but i hit LINEARIZE, is the tagged/interpreted footage converted? or are the values (if indeed gamma encoded) being read as linear 'incorrectly'? Should I impose a conversion on an adjustment layer? Or should I not linearize my workspace at all?

Also, when I plan for more than one output, usually here it's for web (srgb) or tv (SDTV/REC709) or the theatre in sofil (DCDM) or maybe one day if I live to see it, a film print, do I grade accordingly via adjustment layer, or do I let the module do its thang? If I let the module do its thang, things look different on my screen naturally, but will they be fine on the output?

So, my question in simple form is, I have a project. For three outputs possibly, do I work in a linear space? Do I leave footage inside as it is with Color management turned on? and when and why do I use the converters in utility? There's more but I need to get this a step at a time.

I understand the theory so far, the graphs I'm getting to know. The actual what happens when I do this or that I know, but it stills seems that there's no set protocol, so I just need to know why I'm linearizing. Again I don't composite, I do grade though, and I do apply effects that do things to the look of the footage.

I dunno I know it seem sin the air, but I need the steps to working a project in ae cs5 for these outputs (hd, web, dcdm, film print) keeping in mind my footage sources (dslr, ex1/3, red, alexa).

Why do I linearize?
Do I leave footage as is? Is it converted to my space once I tag it with its profile?
Do I grade for exports separately in different output comps and match by eye?

T
 
Most of your questions I can't answer because I don't have the proper context. That is, I'm not sure what you mean when you say "linear record", nor do I know in what program you "hit LINEARIZE". Still...

One way to look at it is that the data in the file has no meaning until it's translated into visual (hue, saturation, and luminosity) and displayed with an output device. Then, the output device has to conform to it's own specifications. Just because you assign a profile like REC.709 to a TV doesn't mean the TV will actually conform to REC.709 -- to do that, one must calibrate the TV to meet REC.709. The problem is that the factory calibration is often cheap, quick, but not very through. And TVs age. So for best results they should be calibrated on occasion -- typically with a spectrophotometer, software to drive it through some tests, and a technician with access to the service menus to tweak the hardware to bring the TV into compliance with REC.709. And part of this is linearizing the TV -- particularly the neutral axis (grays), from black to white.

The above applies to all output devices -- computer monitors, production monitors, to film release print burners. This is why you find some computer monitors with "a puck" like the NEC Spectraview system, why production monitors have the controls easily accessible and color bars are used to bring it into tolerance fairly quickly, and why they burn and process control strips when making release prints.

As to cameras, often the camera will embed a working space into the file. Typically this is for the main target output, such as REC.709 for HD TVs. If you view it on an NLE workstation with a computer monitor that can display REC.709 (such as the EIZO CG241W) you'll see exactly what it's going to look like on a HD TV. If you're going for a web release, you might want to convert the profile to AdobeRGB or SRGB, and set the monitor to display the correct working space.

For workflow, you'll typically want to edit and do your major color grading in the working space that gives the highest gamut (typically for film release prints). It's easier to down grade your output, so you start with the biggest gamut. Then if you need to make a release to a smaller format (such as HDTV (blu-ray), or even SDTV (DVD)) you regrade (this is what wave form monitors are used for, among other things) for these lesser formats.

I'm rambling now and not making much sense to even myself. If you have more specific questions, try asking them in the Adobe subforum here -- you'll get more answers because more people will be looking at that subforum than than "special effects / stunts" I'm sure. Plus, many of those people are way better than I am at this stuff, and much more articulate to boot! :shocked:
 
Again thanks bruce, but to make my ramblings into a simpler question, and also to check if i'm getting this out of your paragraph as an answer, if the footage comes in tagged with let's say srgb, and i'm working in a rec709 space for a rec709 output, i have to change the embedded profile to match the workspace?

I will hit up the adobe guys shortly. I'm sorry again I coulda sworn I read color in here...
 
You must log in or register to reply here.
Back
Top