Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - agentdark45

Pages: 1 [2] 3 4 ... 38
[Max] Daily Builds / Re: Corona Scatter speed improvements
« on: 2020-11-10, 13:39:45 »
Here is my features list :
Ability to control z offset and scale for each model
Distribution modes different from the standard one (like forest Dense, full, isle, scattered...)
Exclude areas, exclude objects
Clusters mode
Surface direction and slope range

I'll add to this:

1. Mappable distribution of each scattered object (within the same Corona scatter), with collision detection between different objects.

 - Example: I want 4 different grass clump models in a scatter, but want finer control of how each clump is mapped within the scatter. Currently you'd have to create 4 different scatters, and collision detection won't work between the scatters. I've been fudging this using various masks and inverted masks, but it can get complicated quickly.

2. Better scatter object collision detection.

- Example: the generated convex hulls aren't great for items with complex geometry (trees with long branches e.t.c) when collision detection values are higher.
- Example: trying to densely pack the volume of a sphere with objects, whilst avoiding objects clipping the sphere or self intersecting.

3. The ability to delete/manually add objects to the scatter (after they have been scattered).

- Example: You like the general feel of the scattered items, but there's one pesky tree that's blocking your view - I want to remove just that one tree without affecting the rest of the scatter. Sure I could keep increasing the random seed value until I find a more suitable scatter distribution for the view, but that can often be quite tedious.

4. UV mappable colour/reflection control of scattered objects.

- Example: Creating patterned rugs easily. I want to tint certain rug fibers in the scatter with a colour in a well defined pattern.

Gallery / Re: Beer
« on: 2020-10-28, 15:12:50 »

Thanks for the breakdown! I'm a max user so having a bit of trouble getting my head around the material did you merge the separate side/top layers into one object? Is the foam one mesh, with Multi ID material or a layered material? Also what's going on with the Vertex maps here?

I've had a bit of trouble doing this in the past with SSS materials and Corona, it doesn't play nice with multi/layered SSS materials. One thing I'm thinking is whether you could triplanar map the foam (using different bitmaps for the top/sides), that way you could keep everything contained in one single material for the SSS portion. I guess this could also be done using a composite map with different UV channels for the bitmaps?

Would you mind linking me where you got the bitmaps or if there's a possibility to purchase them? Would love to run some tests of my own.

Gallery / Re: Beer
« on: 2020-10-27, 23:42:11 »
Thanks :)
The work is very simple: duplicated geometry, one with displacement facing inwards, the other one with displacement facing outwards.

Wait...wut. Can you show us a breakdown of the foam/material? It's extremely impressive and I'll definitely need to utilise this in some of my projects.

Is the displacement map custom or can be it be obtained anywhere?

The one thing that looks off to me is that when looking at reference pictures of beer foam, it generally "merges" into the wall of the glass, whereas your foam sort of sits inside of it (kind of like when people don't intersect liquids with the glass container - it looks odd).

So I stumbled across this article regarding GPU based path tracing (specifically using CUDA and OptiX):

Some interesting takeaways:

"...the GPU version runs all of the same C++ code as regular pbrt to generate camera rays, compute values for low-discrepancy sampling patterns, evaluate and sample lights and BSDFs, filter image samples—pretty much all of the core rendering computation...

And GPUs nowadays are fast… After a little performance work after HPG, now pbrt on the GPU is even faster than it was when I gave that talk. Speedups versus pbrt running on a 6 core CPU are generally 50-100x. If you feel like a 6 core CPU isn’t a fair baseline, then it’s about 10-20x faster than running pbrt on a 32-core Threadripper 3970X"

There's a video link on the page regarding the coding - this is all well above my paygrade, so maybe the devs could chime in on this? Is this some ground breaking new tech/code, or standard marketing?

Gallery / Re: Nuñez Balboa
« on: 2020-10-21, 01:04:54 »

There is really nothing strange or exceptional about it. The light is simply a cloudy hdri. Much of the softness in the light is from the curtains on the windows. For the curtains I use a rayswitch material, so I control on one hand the visual aspect and on the other hand its effect on the light. This way I have more control than using only 1 material.

Post production is also very simple. (seeing it now I might have even overdone it in some points) As the interior is quite monochrome and where white predominates, I can pass with a render with hardly any contrast (if I had different colours, this would probably have been more difficult). Normally, I just save the render in 16 bits tiff and adjust it in Camera raw. If I'm still not satisfied I adjust the lights and shadows separately with a mask. Something similar to the Sharpness method that I explain a little later.

The renders are at the original size. The only system I use to give sharpness is a high pass layer where I remove the lighter areas to avoid the appearance of halos. It may not look very clear in the pictures, but I recommend you try it
Apart from this, I just give it a little chromatic aberration by deforming the corners of the red channel, a little vignetting and I am done.

Hey thanks for taking the time to post the breakdown. I typically avoid much post processing in photoshop, but I'm going to have to change that after having a look at your results - you really pulled a lot of detail out of the fairly bland VFB image.

Gallery / Re: Nuñez Balboa
« on: 2020-10-19, 01:03:02 »
Wow! These are stunning. One thing that stood out to me is how natural and photographic they are.

Would you mind sharing some insights on your lighting, tonemapping and any post processing that went into these? The black and white levels are absolutely spot on, great dynamic range and no visible black crush.

Also, were the renders downsampled at all? They have a great level of sharpness and softness (if that makes sense?).

Work in Progress/Tests / Re: The Trail to Machu Picchu
« on: 2020-10-15, 14:10:46 »
Nice, definitely interested to see how the water was created. I haven't found a good, quick and resource efficient solution for foam/whitewater yet.

Gallery / Re: Natural interior
« on: 2020-10-08, 23:28:32 »
Thats a real nice boucle material, would you mind letting me know where you got it from (or if it was custom made?).

Two questions:
1. What is your RAM frequency?
2. Wat are you CPU temperatures during rendering?

So I just ran the test again with the Corona Benchmark tool, monitoring with AI Suite. Temp hit 68C and stayed there as the clock speed dropped, albeit more slowly this time, shaving about 20 seconds off the time. Clock speed in Corona registered at 2.9GHz.

My RAM is 3200 but I just looked in BIOS and it shows 2133 - not sure if that's normal? If i try manually overriding to 3200 it gets stuck in a loop when booting and restarting itself. Eventually I managed to get a POST screen and get back in to BIOS to set it back.


You'll be hard pressed to get your RAM over 2933mhz with that CPU, at least not with major subtiming tweaking. I would also monitor your motherboard VRM temperatures, it's definitely a known cause of massive slowdowns (the motherboard will throttle the CPU speed to avoid melting itself).

I do feel like the value proposition of GPU's for consumers (at least in terms of rendering vs single monster CPU's) is highly overlooked, and should be factored into development decisions.

For example, the 3080 is about twice as fast in Blender/Vray GPU/Octane than the previous 2080ti:

Now consider the cost of system upgrades for end users:

One could easily get 4x3080's or two+ 3090's (with 48gb's shared Vram since it's the only model in the range that supports NVLink), and use them in pretty much any barebones system, within reason. Now compare this to someone wanting to upgrade to a Threadripper 3990x based system from an older platform, you're looking at a hell of a lot more money all things considered - and this is coming from a TR 3970x user.

There are times where I use Fstorm and my single 2080ti to render scenes that Corona just can't keep up with (mostly high end product shots/animations with DOF). Others on the Fstorm facebook group are also pumping out stunning high res interior visuals in the 30 min render time region.

Looking at future CPU development, I can't see a 256+ core Threadripper emerging within 5 years (at least for a reasonable price), however if we keep getting double the performance in the GPU space every 1.5-2 years, and a doubling of Vram, out of core tech, dedicated raytracing core utilisation in render engines, and NVlink keeps being updated I have no doubt that GPU based rendering will dominate the market.

[Max] I need help! / Re: Render Time
« on: 2020-09-11, 13:54:40 »
Overall scene albedo plays a big role in lighting composition and render times as well. In the following fringe example, there is an indirect light with fixed intensity.

Image Comparer:
  • On the left image, you will see an interior with 200RGB diffuse material, 0 global EV. finished in 9min 46 sec. at 19.8mil. rays/sec with 1.98% image noise.
  • On the right, an interior with 150RGB diffuse material, 3.5 global EV. finished in 18min. at 19mil. rays/sec with 2% image noise.

For the second example with an interior hidden wall light (direct), same materials.

Image Comparer:
  • On the left image, you will see an interior with 200RGB diffuse material, 0 global EV. finished in 49 sec. at 23.3mil. rays/sec with 3.55% image noise.
  • On the right, an interior with 150RGB diffuse material, 2.0 global EV. finished in 1.2min. at 22.6mil. rays/sec with 3.93% image noise.

Very interesting tests!

This has probably been discussed a million times before, but for a typical white wall is the "stick to 200rgb" rule still sound? What are the upper limit rbg diffuse values before things start going beyond reality?

Agree. I often need to go back to the "old" displacement method, because 2.5D gives to many artifacts. Sometimes tesselation before displacement helps, but not always.

Would be great to see a fix for this.


I ran into this issue the other day, it occurred on a curved wall that had a fine ribbed wavy displacement map applied to it. 2.5D displacement produced massive "spiky" black artefacts so had to go back to the old method with 1px displacement resolution...needless to say RAM usage and pre-computation time went through the roof. Increasing the tessellation of the curved wall didn't help in removing the artefacts.

Gallery / Re: Essence - 4K Animation
« on: 2020-07-21, 19:37:45 »
We have tested on serval systems. Can you specify the shot where jitter happens? If anyone else is experiencing the same please do reply. Just to make sure it's not technical playback issue in 4K.

Hmm, so the judder issue appears to be Chrome related - watching the same video at any resolution in Edge (bluergh) and it's buttery smooth! Looks like a totally different animation now :)

Great job again!

Any tips for optimising render time/denoising settings? I've yet to attempt a proper interior animation (even with my Threadripper 3970x)

Gallery / Re: Essence - 4K Animation
« on: 2020-07-20, 21:05:13 »
Nice animation, but some of those shots look quite "juddery". Is this a frame rate issue or a lack of motion blur?

Super craftmanship :- ). Fills me with anxiety only thinking of eventual maintenance.

10/10 for choice of tubing and fittings!

Yep, not looking forward to that!

The 3995wx seems like an odd move from AMD considering that the Zen 3 consumer launch is months away...and Zen 3 TR's will be out towards the end of the year.

Pages: 1 [2] 3 4 ... 38