So, once again I appear to have been bitten by a less than ideal default setting. In this case, the default in question is Linear Workflow being enabled. I have created a bunch of projects and rendered them out over the weekend as 8-bit tiff sequences (psd layers), straight alpha. While all my test renders looked great, bringing in the finished project into After Effects yields objects with translucent materials being nearly invisible.
If I have to re-render all of these, I’m going to want to shoot myself (as just switching off lwf will likely result in me subsequently having to rework all the materials) ... please tell me there is some way I can get After Effects to see these renders “correctly”.
Hi Will,
I do not understand, you use the linear workflow (needs Float if the pipeline is supposed to work in linear light) and then (?) render in 8bit/c integer ... which is a gamma based format. This is certainly not at all how it should work in the first place, nor make that any sense.
As I have no idea what you have set up, do you mind to share a single frame from the rendering? Maybe I can find some “Last-Aid”.
Right, well after mucking about in AE, I was able, I think to solve the problem by switching the workspace to the profile that is embedded on my renders and checking the ‘Linearize Worklfow’ checkbox in the project settings, keeping my fingers crossed that this doesn’t cause other problems 8-bit lwf files can cause.
Well, I hadn’t intended to work in linear. I usually don’t, but this install of C4D is a new one at my new job, and I had forgotten that the setting is on by default.
I understand. Yes, Color profiles can mess with “your” work a lot, but also if they are applied to things that shouldn’t have any profile “attention”.
I haven’t rendered in 8 or 16bit/c integer since years, as I love to have things adjustable.
However, you need to check if the alfa works as expected, as I have no data from your side, render a single frame in non linear and compare the results. So you know if it works after some adjustments, like keeping “RGB values intact” etc..
Understood.. will do. I actually only rendered on alpha in the event the producer decides he wants a background other than black. Thanks for the link, I’ll have a look at it in a few.
BTW - I do really like the adjustability 32bit can offer, but I know it takes a ton of extra disk space, which can be a concern, but does it also increase render times?
Hi Will,
C4D renders in 32bit/c float anyway. So no longer time. Disk space, well, I just ordered a new disk 2TB for $114, take that times two for a 1:1 back up, and I do not see any reason to stick with non float. Seriously.
Just save out an OpenEXR file (Do not use the Quicktime versions…it might normalize the results, not nice) and compare the file-size. Then think about the options to adjust, especially color grading with 8bit/c is painful.
In 2009 there was a problem with EXR files and I had to use Tiff for a while, which resulted in three times the amount of hard disk space, which was quite expensive for me, as I rendered in Cinemascope size. I love OpenEXR and push that format since many years, where ever I can. The last company I discussed it heavily was Photomatix some month ago, to have it as default there (instead of [eeeek] the 4*8bit HDR/RGBE, which is not at all a 32bit/c format!), Photomatix 4.2 has that feature now. A great company! As it is one of my main photography tools, I’m happy to not have to set this up each time.
I’m so used to “paint” light into my work, I never want to go back to integer ever.
Interesting, I was just setting up the renders to do over again as 32-bit tiff, when I saw you had posted again… I’ve never used Open EXR, I’m assuming it plays nice with After Effects CS5.5? Any caveats I need to be aware of?
So far I render out single “layers”, if I work in Ae with it, since you can use the “.aec” image sequence are not a problem, but simple to replace. (Fnord has a nice Plug in for that…)
For NUKE I would render in layered formats, as you can then access everything in the channels, and you do not end up with uncounted Read nodes. I mention NUKE here as this is pretty much what Postscript is to Illustrator, nearly.
OpenEXR is written and developed from ILM and widely used in the industry, namely Digital Domain, the initial developers of NUKE.
I have some tutorials in work about linear-light and why it is mandatory for professional work. It will be not a short story, as I dig deep this time, and Charles Poynton’s newest edition (Digital Video and HD) is currently my hottest book to get through all of that (after many others I have to admit). There is so much good as well bad (!) stuff on the web about linear-light workflow, that I want to clean up that field once and forever. To make it save for the people here on Cineversity. From my point of view, there is no way back to 8bit/c ever. :o) It’s not simple after decades of 100% gamma stuff…
Understood. This is a fairly simple project in which multi-passes aren’t really needed, so I should be ok there.
I am still wondering if I need to uncheck the 8-Bit Dithering option in the render settings? It seems like it should have grayed out when I picked the Open EXR setting which only has a 32 bit option. The manual makes no correlation between these 2 settings, but I want to make 100% sure, as I can’t afford another waster round of renders.
Short answer is no, that will not affect any 32 bit results.
This option (Dither) is only needed to overcome the shortcomings from 8bit. It sounds normally a lot of colors that 8bit can produce with three channels, but the devil is in the detail, which means, when you have many similar colors, especially the blue in the sky is such an area. If not dithered, banding might happen. As this “noises-up” the colors, the noise will be more prominent in the moment you need some heavy color correction. Then better no dither in the rendering-output, switch in Ae to at least 16bit/c, do you “color-correction” and Ae automatically dithers it for the 8bit output.
BTW in most areas people love to work with 16bit/c float also know as Half Float. It has a great precision over 30 stops and a little bit less precision in 10 extra stops. But I personally have never touched it, which is maybe a little bit over the top, but I like the idea of never lose anything.
