Mike A

I hope Maxon can keep up the pace for future releases, because they need to. It's good to see the useful updates in S26, and positive to see the start of Redshift integration, and the inclusion of Zbrush in Maxon One at no cost increase - but the big issue remains. What's the progress on moving fully to the new core? Silence. Capsules are currently a stopgap: scene node native performance 'internally', but choked as soon as they are dropped into the object manager. When will we see a fully 'scene node native' object manager and the speed increases that have been promised year after year?

It's an interesting point about CPU vs GPU farms. I do use GPU farms for animation work as required. Currently RS CPU is about 10% of the GPU rendering speed of a typical 'balanced' CPU /GPU workstation with one GPU. Let's say with development they get it up to 20% relative speed. Anyone any thoughts on whether that might offer better value for farm based renders - as opposed to GPU farms? Ie: Look dev and set up on local GPU workstation - main render runs on CPU farm.

I think it's absolutely fair comment that Arnold / Octane / Corona have delivered more beautiful images out of the box than RS. A key reason is the RS developers have had a very strong focus on speed. A lot of the material defaults - glass being a good example - were set up to deliver a 'good enough image at the fastest possible speed' as opposed to the other renderers where the defaults are more focused on quality. Can you deliver 'Arnold quality' images from Redshift? Well, that's a personal judgement of course, and one you need to make in the light of your clients and their requirements, but you can certainly deliver images that are very, very good. On the positive side: 1. Quite a few versions ago RS / Maxon revamped the render settings to have a 'basic' and 'advanced' mode so they are easier to get into for new users. 2. YESTERDAY RS introduced Redshift 3.5 with a new 'Standard Material' - which is 'insipred' by the Arnold Standard Surface and has some significant improvements over the now legacy 'RS Material' in terms of material design, energy conservation, rough surfaces, thin film and SSS. 3. Random walk SSS is expected to be added in the next few months. 4. Volume improvments and other work is also well underway. You can check out recent features and future plans on the official Redshift Trello page at: https://trello.com/b/QASr74yB/redshift Do i think RS is the ultimate renderer for every requirement? No : ) But I've found it capable and reliable for all my work, delivering an image quality that my clients are quite happy with. It might be worth you checking it out!

I can't answer your Arnold questions - minimal experience there. but maybe you should consider Redshift as an additional option? I've plenty of experience with that if you have any questions!

I just tried my standard demo scene above. Rendering at 1920 x 1080 px I get: 2x RTX3090's: 35.8 seconds = relative speed 1.68 1x RTX 3090: 60 seconds = relative speed 1 5950x CPU only: 661 seconds = relative speed 0.09 So 5950x CPU is approximately 9% speed of a single 3090 in this test. Having said that, this is a quick render on the 3090's. A significant chunk of that time is spent preparing and loading the scene into the GPU, which is not relevant for CPU rendering. So I guess on an animation render the CPU would be relatively even slower - maybe nearer 5% than 9%? Your mileage may vary of course 🙂 I'm sure this will improve. Kudos to the Redshift team for getting it working at all!

First informal 'tests' suggest this initial CPU version of Redshift comes in - very approximately - at about 5 to 10% of the speed of a single RTX3080 when using a recent CPU (eg: Threadripper, Ryzen 9 etc). The key question: Is it faster than Physical : )

How do they manage to make the demo scenes so bad? Did Dave McG knock them up in his lunchbreak?

ROFL... Did someone get mixed up between EST and PST time zones....

1. The feature nobody expected: "Zshift" 2. The feature they've sneakily removed without telling anyone: "C4D"

1. The feature nobody expected, and nobody wants: 'Cerbera mode' - mouse hand electric shocks for tris, complex poles or the unspeakable N... 2. The feature they've sneakily removed without telling anyone: Cube.

1. The feature nobody expected, and nobody wants: Full Blender emulation mode. 2. The feature they've sneakily removed without telling anyone: Mograph - it was incompatable with the 'new core'.

Every year around release time we have a wishlist thread. This year I'd like to encourage your nominations for the 🍓 C4D Golden Raspberry Awards 🍓 in two categories: 1. The feature nobody expected, and nobody wants: 2. The feature they've sneakily removed without telling anyone: Kudos points for the most accurate and for the most creative suggestions : ) - and let's keep it light : )

Agreed 100%. Impressive stuff.

I imagine that Blender 'Correct Face Attributes' has it's limitations, but for a well contructed UV unwrap it looks very handy...

Nice work Matt and Arte. I have been tempted by both Corona and Arnold - but I've stuck with Redshift so far. It's easy to believe 'the grass is greener on the other side', but as Arte's comments confirm, there are pros and cons on each. The renders do look good though 🙂

As long as it can't put a red cube on top of a blue cube, we're safe. But once it can do a chrome sphere on a checkerboard floor...

It looks like you got the answer : )

Try this: 1. Render your bottle cap with a square plane in place of the label. The plane should just encompass the circular label, and be positioned on the cap surface. This image will be used for reference only. 2. Create a square Photoshop document and create your circular label with no perspective. Save as 'label'. 3. Open your (unlabelled) bottle cap image in Photoshop. Add the reference image created in step 1 above it by using File menu > Place embedded. Add the 'label' document as another layer on top of those. 4. With that label layer selected, select: File menu > Free transform. Now right click on the layer in the editor window and choose 'Distort' from the menu. That will enable you to move each corner of the box freely. Move each corner of the box to match your reference plane in the image. Your label should now be in the correct perspective. 5. Switch off the visibility of the reference layer. To update / edit the label simply double click on the label layer smart object, edit it and save. The perspective distorted version will be updated.

Thanks for posting Pinin. That Bernini sculpture is quite astonishing.

I've been shooting images for decades - mainly amateur but some pro work as well. For years we've wanted softer softboxes with smoother diffusion and have cursed the unsightly umbrella rib lines in our reflections... and now in the world of CGI we're free from those flaws. Yay! But, umm... somehow it doesn't look right LOL!

A very interesting thread : )

Fabulous stuff Evan. I enjoyed the renders, and I'm a fan of the danish design aesthetic too : )

The first 10 minutes of his Video from Adrian - the Rs tech support Guru - will help you. It is in Maya - but the same procedure / approach translates directly to C4D. https://vimeo.com/231615490?embedded=true&source=vimeo_logo&owner=2104016

As Daniel suggests, it's likely the ranges set in your Rs Volume material are not appropriate. I've worked with volumes in Rs quote a bit - both VDB's and native C4D generated fog volumes and both work well - but you do need to find the right ranges for the RS Volume material. As there are no standard 'value ranges' for these numbers they can vary over orders of magnitude. If you need to enter very small values - ie less than 0.001, or very large values, you can use the Advanced tab of the RS Volume material to remap those ranges to more manageable numbers. In the advanced tab try setting the 'Old Max' values to a higher values - 100, 1000 or more - then readjust both the Scatter and Absorption vales in the main tab. If you're able / willing to upload your VDB it's easier to help.

Far too philosophical 🤣