Don’t do VFX work on graded footage. period. But.. but.. LOG profiles, color spaces, color science, gamma curves, RAW, transforms… Gosh, I hear ya. It’s really tempting to skip the headache of color spaces when working on VFX productions with professional video production houses.
Nonetheless, from our experience, it’s always worth working on the RAW LOG files. Avoiding working VFX on graded footage will help your production for a few important reasons.
- Maintaining Dynamic Range
- Pixel Integrity and information
- CG production consistency
1. Maintaining Dynamic Range
Starting with the idea that high-end productions shoot on quality cameras for many reasons! One of them is for capturing a high dynamic range; Details and information in the highlights, and shadows. All this extra information in the image provides creative flexibility in post-production.
With that in mind, handing over color graded images to the VFX house usually means a premature creative destruction of information. Limiting the creative control very early on in the pipeline. Which really doesn’t do the production justice!
Example of RAW vs ‘Graded’
Left: RAW footage in LOG, color transformed in NUKE
Right: Same footage, but with the transformations baked into the image
For example here, you can see how much information would be lost in image on the right. Pay attention to the highlights in the sky and on the wooden table. When we dim the image, you can see that there is much more detail and color in the LOG file.
Some VFX context: say we rendered a CG metal item and placed it on the table. The reflections on this metal item would be severely compromised if it want it reflect the sky from the image plate.
2. Pixel Integrity and Information
Following the mantra of maintaining data in the dynamic range, we want to also maintain pixel data integrity! During the grading process and export, pixel values and data integrity may be crushed or stretched to their limits. So whilst the export as a whole may look perfect, these pixels might be prone to artefacts later on.
Final image compression should take place at the end when the video is delivered to the client. Doing VFX on graded footage can potentially limit the flexibility and overall output quality throughout the pipeline when pixel data is crushed before handing it over.
Example of RAW vs ‘Graded’
Left: RAW footage in LOG, color transformed in NUKE
Right: Same footage, but with the transformations baked into the image
Let’s call this the ‘resting’ form. Where both images essentially look identical.
Here we have applied a heavy saturation boost to both images, illustrating the effects of poor pixel integrity artefacting. You can see that although in the ‘resting form’, both images look identical, the image on the right starts to break into blocks when pushed.
Zooming in to a more detailed comparison, we can see the the image on the right really breaking into blocks of color artefacts. Maintaining pixel data integrity from the RAW LOG file vs a compressed ‘graded’ file allows the colorist to push the image further without ‘breaking’. Even after CG and VFX elements are composited on to the image.
Image compression is for the end, not the start of the production
Unfortunately, if we work on an already compressed graded file and deliver it back to the colorist, the flexibility to push or pull this image is severely handicapped should the project require tweaks to this shot later on.
3. CG production consistency
Finally, apart from maintaining data integrity and dynamic range, working in CG for a VFX production also relies on a lot of real-world understanding of color and values. Baking in color-transforms and grades before using these footages in the CG pipeline muddles up these values and forces the artist/software to compensate for the color grade/transform.
Example of RAW vs ‘Graded’
Left: RAW footage in LOG, color transformed in NUKE
Right: Same footage, but with the color grade baked into the image
Even in the very unlikely scenario where the client and production house have ‘locked’ in the color grade (meaning no more changes), it’s extremely difficult to work on the graded footage due to the variety of color variations a particular object can have after a color grade. — Blue shadows, purple shadows, orange shadows. Lot’s of things that would require lots of magic to physically reproduce faithfully.
A lot of the work done within the VFX pipeline is “re-creating the real world”—in CG. So in the same way a video production house doesn’t physically paint/light everything on set to match a color-grade 1:1, the CG scene should not be ‘physically’ altered to match a color grade either.
Looking at the image on the right, you can imagine how challenging and inaccurate it would be to work off the color graded image, versus the RAW image on the left.
Project example: Jasmine Sokko — Tired
Here’s an example of a project where we worked from the RAW footage, and delivered back the footage in the same colorspace so the colorist can continue to work on the footage as before, but now with a giant CG hologram cat. :3 — See project here
RAW anamorphic footage in ARRI v3 Log C Color transformed in Nuke
(ARRI v3 Log C >> Linear sRGB)
Composite of 3D renders
Color transformed in Nuke, back to ARRI Log C.
You can see that through this workflow, the delivery from VFX house (us) to production house is essentially the same footage, in the same color space that respects the pixel information as much as possible. This allows a lot of creative freedom for the artists that work on our files down the line.
Now that we’re all caught up on why we don’t do VFX on graded footage, and working in RAW/Log is the creative’s galore, we will be sharing our VFX workflows when working with professional video production houses in the next post!
