First of all, thanks for the amazing release of houdini 20. As always, it's a big step forward, not only with Solaris/Karma, but with all the amazing new features and updates!
However, there are some aspects that we find lacking in Solaris, and would like to know some thoughts on that.
1. my biggest complain right now, is the way how you iterate look development, or sampling quality in Karma. In short, why doesn't it work more similarly to Mantra render view? Why do i have to have at least two windows open all the time? Viewport for clicking the render button and taking snapshots, and then render gallery for actually looking at it? The functionality of these two windows is overlapping quite a bit, and it is a huge waste of screen space in my opinion. Why is it not possible to run renders from render gallery window (using existing camera)? To me, that seems like such an obvious thing to be able to do. Or not? Ability of being able to move freely in the viewport, and see the render updating almost instantly is cool, but not that useful for us. What we need most importantly is to be able to see final quality image from the camera view, zoom in and out, inspect pixel values and AOVs. All that while the render is running. Using Clones tab kinda allows for that, but it's yet another unnecessary window - all we need is a render button in the render gallery window.
2. Karma pixel/ray sampling - it's got much better from times of H19.0, both in speed and settings. I love how instantly i can see the first pixels. However, in my tests i frequently ended up in situations where i could get presentable quality render in 1 minute, but then sampling it to the final quality was kinda problem. Especially emissive materials are an issue, as that can only be cleaned up by primary samples, if I'm not mistaken (treating is as light doesn't seem to speed the render up).
3. Following up on previous point, there is something in Redshift I love, and it'd be awesome if it was possible to do for Karma too. Talking about automatic sampling. You only have ONE slider that controls amount of noise in the image. That's it, and it works like a charm. All the calculations regarding number of samples for individual lobes are handled by the rederer "for free", and the resulting render is always faster and cleaner than when i try dialing it manually. I admit in the last few years i became super spoiled by it, and forgot how time consuming it may be to get the right combo of numbers for noise free images. Now, when i know this is possible, I want it in Karma too
thank you, D.
