I can’t speak for everyone, but here are a few of my pet peeves:
I am afraid I would not be able to fund that development alone given how expensive competent programmers are. But if you made something like Patreon account, containing a list of features with specified funding goals, I’d be frequent supporter, and I am sure it would motivate quite a few more people.
The issue with funding Blender development is that it’s somewhat vague. This kind of model would allow people to know exactly what they are getting, but at the same time, I can imagine it being quite tricky and annoying to manage.
I’ve been following the first thread, it has been quite educational.
I have been considering better ways that people can back their feature requests with money, since that is a largely non-existent job description, “freelance developer for open source software”. I wouldn’t expect any individual to be able to fund a feature, but voting with money is much more likely to get actual development when it comes to this sort of thing.
What would you expect a feature such as your second link would cost? This is just out of interest at the moment, because I’m unlikely to be able to change my work situation in the near future.
I don’t think I am qualified to do the estimates, as I don’t code. I am just an artist. But let’s try shooting from the hip:
Second feature is really just an extension of principled BSDF with a functionality that is already possible to do with nodes, just unnecessarily complicated. I can even do it and wrap it in a node group, but it’d be messy. It’s about internally adding thin diffuse transmission BSDF already available as a separate node into Principled material and making introducing a switch between it and the actual solid subsurface scattering, and making them share the input slots.
I’d say about 2-3 days of work to get that done + additional 2 to implement review feedback from BF members to get it into a master. At a dignifying rate of $400/day that’s $2000 for that feature.
The ShadowCatcher one, that’s a lot more work. I’d say at least 3 weeks, or 15 days, so $6000.
I have to admit these numbers are quite high, somewhat reducing the chances of successful ending, but at the same time, it’s inappropriate to expect programmers to work for peanuts. Especially computer graphics programmers, as they are usually very few and very skilled.
I think your estimates are reasonable, but I do feel like you might be in the minority who have a reasonable expectation of dev-time and hourly rates.
It would of course be up to the developers to determine what price they would want in order to implement something, and yes, maybe to some, $6000 for a relatively minor feature of a free piece of software, that might seem like a lot. It would be interesting to see if there are sufficient professional artists/studios using Blender who would be willing to pay for things to be implemented for them.
Yup, that’s why I think something like Patreon would be a good test to see if it’s feasible. Even if it failed, only thing that would be lost would be about 30 minutes setting up the account and writing a description for the goal
Indeed. I’ll keep it in mind if my work situation changes.
I do plan to get more familiar with Cycles as a whole in my spare time, so maybe one day I would be comfortable asking for money in order to develop it. That day certainly hasn’t come yet.
LuxRender, Indigo, and Octane are all spectral renderers. Octane is faster than Cycles generally, so there is probably a lot you can do to speed things up.
I think Octane has made some optimisations which Cycles has decided against using, but these decisions might be revised in the future.
There certainly are many optimisations that can be made in a spectral renderer.
I think Octane has a few sampling techniques which Cycles doesn’t have either, but the tricky bit is that a few of these optimizations are patented and can’t be used in an open source engine (so we can’t copy from them and must find another way).
As for what kind of performance impact I would take for spectral rendering, it really depends on how big of a realism bump is possible and whether there’s larger optimizations in the roadmap for Cycles.
For ma as a small freelancer working mostly for industrial companies those ‘effects’ are pretty important. Making a lot of metal things with focus on details forces me to use some nasty hacks and faking a lot of materials hoping that they can match real world examples enough.
Well, for instance my clients are able to see differences between colors of heated iron or other effects with thin film layer involvement that are not ‘correct’. So for me paying 300 euro more, but only once, is kind a crucial.
I think in the same manner as CEO of nVidia (from RTX presentation), even that we have more processing power than ever, production of movies (animated or with VFX) is constantly getting longer because artists wants more and more realism in there. So probably even if for small studios or freelancers (as myself) quality will be more important than speed itself. Even if slightly. And even if for my own satisfaction. I personally prefer to make good things, not only ones that can be sold.
appleseed, Thea, Maxwell, Arion, Ocean… are also spectral, but LuxCore 2 is not among those anymore.
For proper dynamic light transport simulations this is a must.
& Some talks here remind me of the blindness in the past, alike “Who needs: more memory, path tracers, gigahertz’s, GPUs, helmets, airbags, medicine, insurance, mobile phones, digital watches & cameras, flat screen TVs…”.
Some render engines use internally 3 distinct wavelengths for (R,G,B) and corresponding IOR values for transparent materials in part of their codepaths for the evaluation of dispersion (LuxCore, Iray). But they are not fully spectral in the sense that these engines ( the spectral ones) use a whole band of wavelengths (e.g. 31 in case of appleseed) which are used in the evaluation of the surface and volume shader.
Spectral rendering could bring some of those features automatically.
Things like thin film interference, diffraction, dispersion, Tyndall effect, and fluorescence would all be much easier to achieve if cycles were spectral.
Am i blind or have they forgotten to put the Render Seed Option for Cycles back into the 2.8 UI?
Can’t seem to find it.
Oh, and when they put it back it would be nice if it’s animated by default!
