Welcome to our community

Be a part of something great, join today!

  • 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 :)

RED VFX Workflow Discussion ...

PatrickFaith

Well-known member
Joined
Nov 4, 2011
Messages
2,560
Reaction score
0
Points
36
Location
California
Website
www.pudls.com
I'm wondering if anyone is interested in discussing red vfx workflow? I think RED is the best system for vfx workflow, and I've seen some great work in these discussions - so I know there is a lot of expertise on the subject here. As way as organizing the discussion, i'm thinking would be nice to state the tools you are using then the current problem your working on or have solved.
----
To start is off, my tool set is: nuke7,modo601,realflow,adobe suite, python/blender,davinci.

My current problem is i received about 400 plates where the textures appear to be color graded with saturation turned up(i'm stuck with these plates and can't get the plates in raw). I'm sure someone wanted their pictures to be perty. But i need to use the the plates in modo renders, for compositing in nuke. The problem is the plates were color graded with a lot of saturation - so the linear render look off. [ Lesson learned btw is to do all the photography on RED, and require all the plates given over to vfx in r3d ] But for now, what i think I need to do is look for shots in the normal movie that have similar colors, then color match the plate to the redlinear shots, and do this on the textures before render?

It's a lot of work, i guess i could "attempt" to color the render in nuke(or even later in davinci), but i'm thinking it would be better to just bite the bullet and color grade all the plates before render ... but i haven't ever color graded that many plates (i really don't know how to color grade in photoshop ... i guess i could color grade in modo ?). so i guess my question is, when you get plates that have been color graded before they have gone to vfx and have to eventually match with redlogfilm/log2lin in nuke ... where's the best way and place to do that grade?
 
Sometimes I'll just put a desaturation node in my shader. But usually I'll save out a desaturated photoshop.

If every texture though is over-saturated then you can just desaturate the render. Although it does mean you'll need to over-saturate your light colors.
 
Sometimes I'll just put a desaturation node in my shader. But usually I'll save out a desaturated photoshop.
If every texture though is over-saturated then you can just desaturate the render. Although it does mean you'll need to over-saturate your light colors.
Thanks Gavin, I will try all three ... probably depends on the circumstance, i like the idea of desaturation node on the shader ... that will be the easiest for me and will save me a ton of time.
 
My current problem is i received about 400 plates where the textures appear to be color graded with saturation turned up(i'm stuck with these plates and can't get the plates in raw).
You can't just ask the company to send you the plates without correction, or with a less-saturated correction? There should be a post supervisor or a VFX supervisor in charge of details like this.
 
You can't just ask the company to send you the plates without correction, or with a less-saturated correction? There should be a post supervisor or a VFX supervisor in charge of details like this.
Your of course right Marc, normally the studio/production company deals with that type of stuff and i didn't even realize i had a problem till it went to compositing. On my next request I'll make sure i'm more specific, i'm trying not to be the neophyte total jerk new director and am just learning industry norms of what I should be asking at different price points for small indie vfx houses. Orly's suggestion of modify the saturation via ray tracing rendering nodes is working good, at least for this section ;)
 
"RED is the best system for vfx workflow" ...
Without any DOUBT
WORD !

Next question which i haven't really seen any one state is the formula in nuke and ae for max pixel width. Basically what I do for my nuke & ae nodes/composites is:
node_width = ( 1- ( (redcode#-3)*5+20)/100 ) * 4096

so if i'm shooting at redcode 6 and at 4k, my max node width should be 2662.

Anything more on pixels, i'm getting some spatial sliding when tracking/camera solutions. As a money guy leaning over shoulders and not trying to irritate people, I've noticed the only time i see someone tracking at full 5k/4k is when they are doing the tracking points by hand, normally for some wierd skin effect.

Because of this, i've just standardized all my nodes to full container dcp for nuke & vfx (i.e. 2048x1080). What is the node/composite pixel size you guys are using for 99% of the vfx work?
 
Without any DOUBT

I don't know, haven't gotten any footage yet but F65's dynamic range and lack of rolling shutter issues on paper sounds nice.
 
what exactly do you mean by max node width?
and also why would the redcode have a factor is determining the max pixel width?
In AE, i would use two comps: comp_4096to2662 (where it only does a 4k size fit to the 2662 pixel wide comp[this also speeds things up with the cache]) and another comp_2662_track(where one layer is comp_4096to2662, and the track occurs in this comp). By max i mean if the comp is any bigger then 2662, for example it was 4096, and the tracking square was the same amount of pixels as 2662, then larger comp will have a larger tracking error. As for why redcode effects max pixel width, i am not exactly sure, my assumption is it is doing lossy compression on the spatial distances but not lossy compression on the color space, so most of the compression is a result of a loss of spatial information - which the eye really can't see but it messes up pure mathematical functions. So i think they have made redcode so the compression numbers don't effect much from a human eyes viewpoint, but actual spatial information is lost when the redcode compression number is increased. A different way of thinking about this, is if you wanted to do seamless vfx at 4k, you need to shoot at around 6k with redcode not greater then 6:1(or 5k at 3:1).
 
Last edited:
I don't know, haven't gotten any footage yet but F65's dynamic range and lack of rolling shutter issues on paper sounds nice.
VFX on sports type of things would seem to be the real nice feature for a global shutter. There are some nodes now in nuke that will remove rolling shutter, i haven't tried them (point here is why spend a ton of money in the camera if the rolling shutter issue can be handled 99% of the cases in post).
 
Your of course right Marc, normally the studio/production company deals with that type of stuff and i didn't even realize i had a problem till it went to compositing.
My advice: tell them right now there's an issue and ask how they'd like to solve it. The problem is, if you "fix" what you think is a problem, then come to find out that they wanted the entire movie to look like that, now all your effects shots will look wrong when they're cut back into the movie.

Communication is often a far bigger problem to deal with than compositing, rates, or schedules.
 
My advice: tell them right now there's an issue and ask how they'd like to solve it. The problem is, if you "fix" what you think is a problem, then come to find out that they wanted the entire movie to look like that, now all your effects shots will look wrong when they're cut back into the movie.

Communication is often a far bigger problem to deal with than compositing, rates, or schedules.

Agreed, got it worked out ;)

BTW ... big license drop for the next month on MODO and NUKE ... like 40% off ... so is a good time to cut over to NUKE. Also nuke 7 is out of NDA and is public. I really like nukex7, has great red support.

(the nuke discount code is ... NUKE4LUX )
 
Back
Top