Let's start with the fact that he's mixing up several different concepts
- Dynamic distributed processing of isolated packages. (e.g., SETI@home with tens of thousands of active systems, ranging from general desktops to powerful servers that have resources to spare during the night)
- Network rendering
- Cloud-based render services (common interface and management for a homogenous group of render nodes)
- Heterogenous loose networks of amateur hardware
- ...
Then there's the software/version/license issue.
Professional environments either have a streamlined standard or a standardized production pipeline where content is created and then converted for use down the road (e.g., modeling/surfacing in one application, refining and texturing it in another and then making the materials production-ready in a third application whose format is a standard for the later steps).
The version issue relates to how some features can often work slightly different in one version than another and in some patch versions they're simply temporarily broken (happens with many 3D programs), meaning that you can get flickering elements if your resulting render if one frame was rendered in an older and one in an improved or slightly faulty version because version A does something about different than version B (global illumination, depth of field, motion blur, irradiance, ...). How do you plan to coordinate the version update cycles across project contributors on possibly several continents to keep such differences to a minimum?
Blender is free, but it also comes with an atrocious user interface that has contributed to the fact that it remains a niche product no matter how powerful the stuff under the hood has become. If you decide on something else for the rendering part of the pipeline you'll have to consider that some applications also require a paid license for each render node, not just for the machines where an actual human is doing the work.
And don't forget that the renderer itself is only part of the equation. Most advanced renderers nowadays come with their own advanced shaders that won't work with others (e.g., the omnipresent mia_material shader won't work in most other renderers).
Then how do you plan to manage and distribute the resources? Some nodes will simply go down, others will take forever to finish a frame and what happens if you have to distribute more than one sequence or need to re-prioritize during processing because Starship XYZ needs certain content in time for an interview opportunity that has popped up?
Long story short: What he's proposing can work in projects that have started out with that process in mind (like the Blender movie projects), but in most low-budget cases you'd be setting yourself up for failure. As nice as the idea sounds, there's a reason why most professional projects either do this kind of thing in a controlled local environment or at a service provider that has a crapload of identical setups and a management backbone into which they've sunk a ton of R&D investments.
I understand the need to setup controls. My references were simply ideas being put out there.
This isn't something that can done over night. The idea would be to start small, set standards and parameters, then if that works, move from there. If someone who has experience in this can suggest workflow paths using free and/or opensource software, that would be a great, or adapt workflow paths from commercial tools to free/open source.
With powerful free software like Fusion, Resolve Light, Lightworks, GIMP, Blender, Lux Render, Sketchup Maker et al. Along with free render control software. I'm also sure some basic programming will be involved. I'm teaching myself python at the moment.
I don't see why all of us could not get together and come up with a system. Test it, address problems, test again...until a reasonable and easy to implement wide area VFX workflow that meets the needs of low budget scifi movie makers can be realized.
None of this was possible just a couple of years ago.
if we can save just 30% of the time needed to render the sort of vfx shots most of those here are looking for, then it's worth the effort.
Instead of getting snarky with each other, I'm hoping people will lay down differences, lend their experience and talent and create something that benefits all of us.
I'm willing to donate computers for this. I have 3 laptops that would fit the bill from low end to work station. Currently building a new desktop based on 5th gen Intel CPUs. Plenty of Trek Blender models out there to test with, plus terrain and city and plant models. I'm hoping there are others out there that can pitch in too.
Open to any and all ideas on how to move this forward.
