Vray lighting (Terrrible)

Why is the lighting in Vray 5 so terrible? Vray 3 out of the box a user hadn’t needed to do much of anything. Vray 5 whites are washed out and over-exposed.

Rhino 7 - Arctic w Outline, beautiful.

Vray 5 (before manual corrections)

Curves (absurd manual adjustments)

Lens effects (none)

White balance

Final export with minor corrections and yet the lighting/shadows are still terrible.

Yep. It’s terrible. Sun is stuck N @ 90° while rendering in Rhino 7 w/vray 5, it doesn’t matter what position I set it manually in Rhino or Vray.

I’m unsure what the problem could be.

Ran the same rendering last night into the late morning, the sun was still broken. I stopped the rendering as it was taking 10 plus [cpu+gpu] hours and while saving the file rhino crashed.

A new ‘basic’ file of a simple rectangle the box is horribly emissive. The sun shadow seems to move, but the lighting is terrible.

Running again with the head file with GPU only. Still, shadows and sun are still broken.

I’ve reset my bios, removed my teslas, quadros, gtx cards and reset them. Also reset PCIe lane tech from Auto to Gen3. Additiaonlly updated to Toolkit 11.2.1 (461.09)

Hi,
I don’t use Vray for rhino but I use vray in max. I’m surprised no one has an answer for you.
It looks like it might be a camera exposure setting, are you using the vray camera or the rhino camera?
Are you using vray materials or rhino materials on your object/s? Try setting your exposure setting higher or lower and or change shutter speed and iso film type.

I think the native camera in rhino for V7 is much more physically based due to cycles than the V5 camera was.

Sorry if none of this helps as I don’t use vray for rhino but I have had similar problems in max but solved them by using vray materials and the vray camera not the native materials and cameras that ship with max.
RM

The last image posted is very basic and default settings. It has two ‘generic’ materials. The first is a generic white material (rectangle) and the other is generic grey material (infinite plane). The cameras are default. I reinstalled Rhino 7 and Vray 5 and the render is straight from the perspective view. I’ll look into your suggestions… it sounds like that could be the issue but these are default (unmodified) settings that I’m rendering from in vray. I’ve not experienced this issue before.

I often get a black render in perspective view which happens often in Rhino 6/Vray 3.6 and I keep forgetting how I fix that, but that’s an entirely different issue than this. I need to get Rhino7/Vray5 working.

Currently running another head render in Rhino7/Vray5 and I’ll post that here. After that, I’ll try rendering one of my projects from last year again.

box_sun.3dm (117.6 KB)

@bandaid I checked your file and it looks like the custom orientation of the V-Ray control doesn’t work. Maybe @Peter.Chaushev can help or log it as bug.

But you can control the sun position per Rhino sun control. At the Rhino sun control the manual control is enabled. Maybe there is a problem between V-Ray and Rhino, if the manual control is enabled at both. But disabling manual control at the Rhino side doesn’t enable the manual control at the V-Ray side. So, I suppose so it’s bug.

I don’t used the V-Ray sun often and my experience is limited, but per Rhino position control it seems to work like it should work.
The white material is unreal 100% white, the ground is a dark grey and your environment is a studio HDRI. It helps to clean the setup. :wink:

A clean setup:

  • create a sky texture at the V-Ray texture tab
  • copy this sky texture to the environment (paste as instance)
  • use a 80% white for the cube and light grey for the ground
  • set the camera DOF to Rhino camera target (the default 200mm are useless)

The manual mode of the sun of the new created sky works perfect. Here my result:

Here the modified scene:
box_sun (Micha).3dm (118.8 KB)

Your head model scene looks dull because:

  • overexposured
  • white material unreal bright (?)

Also per full GI you get an other look than per ambient occlusion of the arctic display mode. If you like the arctic AO look than you need to work with the “dirt” texture of V-Ray.

3 Likes

Hi, indeed there is a bug in the sun’s rotary widget, or rather, how the data is handled. It manifests itself in certain situations. Anyway that is fixed already and will be available in the upcomming release.
It is also true, that both controls are not yet synchronized. There is a in-house debate whether we should sync them at all (I voted for full sync)

For the time being, please rely on the Rhino’s Sun controls

2 Likes

@Micha It looked dull because… I just realized the new option “Display Correction > Properties > Display corection Exposure (display only)” the other night and neglected depth of field. An, oh duh! Idiot moment. I’ve been using vray3.5 and Rhino 6. I hadn’t updated to Rhino 7/Vray5 until late. In vray3.5 we had “Exposure, Color, Contrast” to adjust. Thanks for the texture tip.

@Nikolay The sun randomly worked with Rhino7 (latest version) in some files, even two/three year-old projects. I tried vray5 sun control but that hadn’t worked at all for me. Rhino 6/vray3.6 everything is fine. Thanks for acknowledging the issue.

In a previous post I mentioned that I was running another high-quality head render, and was going to post it here, well, I was cooking steaks while the render was on its sixth hour of rendering. I went down to the office to check on the render and the computer restarted on its own. Not sure what it was. It wasn’t a system update that’s for certain. Now, my head renders, mind you they’re the same quality/settings, are taking less than an hour… vs 8 to 12 hours or longer. My teslas used to chomp at the bit removing noise and revealing depth. Now after the system crash vray is skipping denoising and running vdenoise with an inferior card than my tesla. Something is wonky.

Really wish ChaosGroup added the ability to save log files for every render attempt, enabled by default. I’d be able to track down exaclty what Vray is doing differently.

Rendering that I just completed took less than an hour. Very poor quality denoise, skipping the tip of the nose and a few other spots. The depth of field before the crash was 5x better than this.

Denoising complete in 167.50 s … not possible.
test_5-K80-0004.txt (17.3 KB)

@Nikolay , @Peter.Chaushev Would be nice for someone from chaos group to address the issues that I presented.

Great the sun works now! Nice work. Stamp is back. :slight_smile:

The Rhino sun will not work if the Vray5 sun is enabled. The sun does not sync with Rhino7 and Vray5 or vice versa as you were debating. I think they should link, but that’s my opinion. I also think there should be an option to save multiple sun positions too.

Version 7
(7.3.21053.23031, 2/22/2021)
Vray
(vray_51001_rhino_win_x64)

My settings with interactive sun

Interactive Sun

Render Sun

Curve’s are acting goofy. Unable to reset curves, they do but not to default points, and unable to delete some curve points. I suspect that there are two static points but they do not reset to default positions.

Curve’s that fall off the plot and unable to modify/adjust even after resetting the template. I have to right-click multiple times to bring up the context menu and doesn’t always appear.

Other issues in the previous post still not addressed. I think the problem was 'Noise threshold goal… %" was not functioning in the previous 3dm file. NVIDIA_AI does not work for Keplar cards so I am having to use Vray denoiser. The problem I was experiencing in the previous 3dm file is the Vray denoiser was falling back on the GTX 970 and not utilizing the CUDA cores on the Keplar card. I’ll test again, but this is the current live situation displayed in the image file.

Additionally, Vray is reporting, not sure if you’re aware of this:
nVidia driver version: 461.40
Checking CUDA driver version: 5001: Installed driver version is newer than the latest verified one!
Number of CUDA devices: 2

@bandaid You can save sun position at the Rhino sun panel. Also you could disable the document sun and should be able to add several own skies&sun.

@Nikolay I have seen the buggy curve control too. After a new point was added to the middle of the curve, the curve can’t be reset anymore (no context menu available anymore). Also I miss to move the handle up/down and not only horizontal. But after I delete the whole curves layer and added a new one than it worked perfect.

image

@Nikolay I can repeat the curves bug: add a new point to the curve and move the handles. OK. Now switch from Rhino to the Web browser, so that Rhino is hidden in the background. Now switch back to Rhino. If I try to move the handles than the handle snap in horizontal position and no context menu is available.

should be able to add several own skies&sun.

Yes, I forgot that was there. Though was thinking of a list, such like named-views for sun positions.

Oh, sorry, I must correct me. I checked it now and see … there is no way to add additional sun lights.

@Nikolay Maybe I have overseen something, but how can I add additional sun lights so that it can be choosen at the sky options? If I remember me right at VfR2 I was able to choose a dir light as sun.

image


Also at the light menu there is no dir light. It can be added per Rhino only, but not at the asset editor.

image

Adding new sun is via SunLight command. The sun is not on the lights list in AE, because multiples suns is a Rhino unique feature, while AE is a shared component.

I will report the curve bug to the core team

@Nikolay Thank you for the infos.

Shouldn’t be the dir light at the list of the AE?

Will all features which are not supported by SU skipped for Rhino?

Clearly ignoring the issues that I’ve mentioned.

@bandaid Which is the open issue?

As stated in multiple posts denoising is not working properly. That is the entire point of this thread was the sun and denoise.

Being a retired systems administrator of fortune 100 companies and also owning my own business I would have canned @Nikolay or whoever else is responsible for providing support.

Completely skips denoising and falls back on the GTX 970, I ran this same file before with identical settings yesterday and the denoiser was running for hours and working as intended. As previously displayed in the screenshots above. Now, the render I just took now (displayed in the progress log below) took less than 5 minutes and ought to have ran for several hours.

The problem is… this is a production issue. I cannot be the only individual experiencing this. It’s affecting business, students, and others. And the response time from support is ridiculously lacking. It took Chaos Group support to address the sun issue SEVEN DAYS AFTER it was reported!

Max. Mpaths per second: 4.777441
Sampling level reached: 2 samples per pixel
V-Ray GPU kernel render time: 73.66s
Total GPU samples 213745744
GPU Memory Stats:
  Device[0] : Tesla K80
    Textures       : 39.30MiB
    Pinned         : 1.13MiB
    Misc           : 1.46MiB
    Internal state : 310.34MiB
    LightCache     : 0.50MiB
    Total          : 352.74MiB
  Device[1] : Tesla K80
    Textures       : 39.30MiB
    Pinned         : 1.13MiB
    Misc           : 1.46MiB
    Internal state : 310.34MiB
    LightCache     : 0.50MiB
    Total          : 352.74MiB
OCLBufman stats:
Owners count     : 12
With replace ptrs: 6
Geom pointers    : 5
Light pointers   : 2
Textures pointers: 7
NVLink allocated : 0
Misc pointers    : 10
Total pointers   : 24
Device 0 statistics:
  GPU time: 36.567636 %
  GPU ray trace time 95.051869 %
  Transfer to GPU time 0.324690 %
  Transfer from GPU time 4.623441 %
Device 1 statistics:
  GPU time: 31.740777 %
  GPU ray trace time 95.113026 %
  Transfer to GPU time 0.272536 %
  Transfer from GPU time 4.614438 %
Number of raycasts: 0 (0.00 per pixel)
  Camera rays: 0 (0.00 per pixel)
  Shadow rays: 0 (0.00 per pixel)
  GI rays: 0 (0.00 per pixel)
  Reflection rays: 0 (0.00 per pixel)
  Refraction rays: 0 (0.00 per pixel)
  Unshaded rays: 0 (0.00 per pixel)
Number of light evaluations: 0 (0.00 per pixel)
Maximum memory usage for resman: 0.00 MB
Number of intersectable primitives: 14
  SD triangles: 12
  MB triangles: 0
  Static primitives: 1
  Moving primitives: 0
  Infinite primitives: 1
  Static hair segments: 0
  Moving hair segments: 0
Peak memory for IPR rendering: 3.76 GiB collected for 22 categories
  Peak memory for "Bitmap": 8.93 MiB
  Peak memory for "Bucket image sampler": 0 B
  Peak memory for "Dynamic displaced geometry": 0 B
  Peak memory for "Dynamic geometry": < 1 MiB
  Peak memory for "Dynamic moving displaced geometry": 0 B
  Peak memory for "Dynamic moving geometry": 0 B
  Peak memory for "GI": < 1 MiB
  Peak memory for "Lights": 0 B
  Peak memory for "Misc.": 155.05 MiB
  Peak memory for "Misc. displaced geometry": 0 B
  Peak memory for "Misc. geometry": < 1 MiB
  Peak memory for "Misc. moving displaced geometry": 0 B
  Peak memory for "Misc. moving geometry": 0 B
  Peak memory for "Progressive image sampler": 1.97 GiB
  Peak memory for "Static displaced geometry": 0 B
  Peak memory for "Static geometry": 1.38 MiB
  Peak memory for "Static hair": 0 B
  Peak memory for "Static moving displaced geometry": 0 B
  Peak memory for "Static moving geometry": 0 B
  Peak memory for "Static moving hair": 0 B
  Peak memory for "Tiled bitmap": 0 B
  Peak memory for "VFB": 1.62 GiB
Maximum memory usage for texman: 0.00 MB
GI engines: 'Brute force' and 'Light cache'
Preparing renderer...
Light tree is enabled with clamp coefficient 0.1 and even split coefficient 0.5.
CUDA renderer requested.
CUDA renderer plugin already loaded.
CUDA renderer plugin instance successfully created.
EXT_RTOPENCL interface obtained successfully from CUDA renderer plugin instance.
Using low thread priority.
Max ray intensity is enabled: rendered result may have incorrect brightness.
Estimated progressive buffer memory usage of 17943 MB is over the limit of 2048 MB. Falling back to per-pixel filtering.
Using Embree ray tracing.
Building Embree static trees took 0 milliseconds, memory used 0.00 MB
Scanning scene for light plugins.
1 lights found.
Using adaptive light evaluation with 8 lights.
Using 0.69 MB for adaptive lights buffer
Initializing CUDA renderer...
nVidia driver version: 461.40
Checking CUDA driver version: 5001: Installed driver version is newer than the latest verified one!
Number of CUDA devices: 2
Initializing environment kernel
Transferring static geometry tree with 12 faces to CUDA devices...
Total number of dynamic triangles: 0.
BUILDING Embree dynamic geometry tree for GPU
Embree dynamic geometry tree contains 0 primitives.
Embree dynamic geometry tree built in 0.06 ms.
Embree BOX tree statistics:
Primitives count: 1 (Mem usage:0.00 MB)
Meshes count: 0 (Mem usage:0.00 MB)
Number of tree nodes: 0
Tree takes 0.00 MB. Build time 0.10 ms.
Scene contains 1 lights
GPU tex loaded sky (idx=0) mip 0, w 765/765, h 765/765, channels 4, 4572KB
Max. Mpaths per second: 4.777441
Sampling level reached: 2 samples per pixel
V-Ray GPU kernel render time: 73.66s
Total GPU samples 213745744
GPU Memory Stats:
  Device[0] : Tesla K80
    Textures       : 39.30MiB
    Pinned         : 1.13MiB
    Misc           : 1.46MiB
    Internal state : 310.34MiB
    LightCache     : 0.50MiB
    Total          : 352.74MiB
  Device[1] : Tesla K80
    Textures       : 39.30MiB
    Pinned         : 1.13MiB
    Misc           : 1.46MiB
    Internal state : 310.34MiB
    LightCache     : 0.50MiB
    Total          : 352.74MiB
OCLBufman stats:
Owners count     : 12
With replace ptrs: 6
Geom pointers    : 5
Light pointers   : 2
Textures pointers: 7
NVLink allocated : 0
Misc pointers    : 10
Total pointers   : 24
Device 0 statistics:
  GPU time: 36.567636 %
  GPU ray trace time 95.051869 %
  Transfer to GPU time 0.324690 %
  Transfer from GPU time 4.623441 %
Device 1 statistics:
  GPU time: 31.740777 %
  GPU ray trace time 95.113026 %
  Transfer to GPU time 0.272536 %
  Transfer from GPU time 4.614438 %
Number of raycasts: 0 (0.00 per pixel)
  Camera rays: 0 (0.00 per pixel)
  Shadow rays: 0 (0.00 per pixel)
  GI rays: 0 (0.00 per pixel)
  Reflection rays: 0 (0.00 per pixel)
  Refraction rays: 0 (0.00 per pixel)
  Unshaded rays: 0 (0.00 per pixel)
Number of light evaluations: 0 (0.00 per pixel)
Maximum memory usage for resman: 0.00 MB
Number of intersectable primitives: 14
  SD triangles: 12
  MB triangles: 0
  Static primitives: 1
  Moving primitives: 0
  Infinite primitives: 1
  Static hair segments: 0
  Moving hair segments: 0
Peak memory for IPR rendering: 3.76 GiB collected for 22 categories
  Peak memory for "Bitmap": 8.93 MiB
  Peak memory for "Bucket image sampler": 0 B
  Peak memory for "Dynamic displaced geometry": 0 B
  Peak memory for "Dynamic geometry": < 1 MiB
  Peak memory for "Dynamic moving displaced geometry": 0 B
  Peak memory for "Dynamic moving geometry": 0 B
  Peak memory for "GI": < 1 MiB
  Peak memory for "Lights": 0 B
  Peak memory for "Misc.": 155.05 MiB
  Peak memory for "Misc. displaced geometry": 0 B
  Peak memory for "Misc. geometry": < 1 MiB
  Peak memory for "Misc. moving displaced geometry": 0 B
  Peak memory for "Misc. moving geometry": 0 B
  Peak memory for "Progressive image sampler": 1.97 GiB
  Peak memory for "Static displaced geometry": 0 B
  Peak memory for "Static geometry": 1.38 MiB
  Peak memory for "Static hair": 0 B
  Peak memory for "Static moving displaced geometry": 0 B
  Peak memory for "Static moving geometry": 0 B
  Peak memory for "Static moving hair": 0 B
  Peak memory for "Tiled bitmap": 0 B
  Peak memory for "VFB": 1.62 GiB
Maximum memory usage for texman: 0.00 MB
GI engines: 'Brute force' and 'Light cache'
Preparing renderer...
Light tree is enabled with clamp coefficient 0.1 and even split coefficient 0.5.
CUDA renderer requested.
CUDA renderer plugin already loaded.
CUDA renderer plugin instance successfully created.
EXT_RTOPENCL interface obtained successfully from CUDA renderer plugin instance.
Using low thread priority.
Max ray intensity is enabled: rendered result may have incorrect brightness.
Estimated progressive buffer memory usage of 17943 MB is over the limit of 2048 MB. Falling back to per-pixel filtering.
Using Embree ray tracing.
Building Embree static trees took 0 milliseconds, memory used 0.00 MB
Scanning scene for light plugins.
1 lights found.
Using adaptive light evaluation with 8 lights.
Using 0.69 MB for adaptive lights buffer
Initializing CUDA renderer...
nVidia driver version: 461.40
Checking CUDA driver version: 5001: Installed driver version is newer than the latest verified one!
Number of CUDA devices: 2
Initializing environment kernel
Transferring static geometry tree with 12 faces to CUDA devices...
Total number of dynamic triangles: 0.
BUILDING Embree dynamic geometry tree for GPU
Embree dynamic geometry tree contains 0 primitives.
Embree dynamic geometry tree built in 0.06 ms.
Embree BOX tree statistics:
Primitives count: 1 (Mem usage:0.00 MB)
Meshes count: 0 (Mem usage:0.00 MB)
Number of tree nodes: 0
Tree takes 0.00 MB. Build time 0.10 ms.
Scene contains 1 lights
GPU tex loaded sky (idx=0) mip 0, w 765/765, h 765/765, channels 4, 4572KB
PTX max register count 64:64
Building optimized CUDA kernel (0) for comp cap 3.7 for device 0...
PTX file generated with CUDA Toolkit v10.0 for CUDA compute capability 3.0.
Optimized CUDA kernel assembled successfully
Total memory for device 0 : 11448 MB, free 10709 MB
Maximum kernel work group size: 1024
PTX max register count 64:64
Building optimized CUDA kernel (0) for comp cap 3.7 for device 1...
PTX file generated with CUDA Toolkit v10.0 for CUDA compute capability 3.0.
PTX max register count 64:64
Building optimized CUDA kernel (0) for comp cap 3.7 for device 0...
PTX file generated with CUDA Toolkit v10.0 for CUDA compute capability 3.0.
Optimized CUDA kernel assembled successfully
Total memory for device 0 : 11448 MB, free 10709 MB
Maximum kernel work group size: 1024
PTX max register count 64:64
Building optimized CUDA kernel (0) for comp cap 3.7 for device 1...
PTX file generated with CUDA Toolkit v10.0 for CUDA compute capability 3.0.
Optimized CUDA kernel assembled successfully
Total memory for device 1 : 11448 MB, free 11011 MB
Maximum kernel work group size: 1024
Running RTEngine
Building light cache.
Tracing 2250000 image samples for light cache in 64 passes.
Using 490.16 MB for light cache preview buffer
Optimized CUDA kernel assembled successfully
Total memory for device 1 : 11448 MB, free 11011 MB
Maximum kernel work group size: 1024
Running RTEngine
Building light cache.
Tracing 2250000 image samples for light cache in 64 passes.
Using 490.16 MB for light cache preview buffer
Merging light cache passes.
Light cache contains 3126 samples.
Light cache base memory usage is 0.8 MB.
Using premultiplied light cache.
Using adaptive lights ver2
Adaptive light grid built in 7.15 ms.
LightCache contains valid adaptive lights data. Building GPU adaptive lights...
Lightmap successfully loaded.
GPU light cache: 4465 samples stored.
GPU adaptive lights v1 built successfully
Merging light cache passes.
Light cache contains 3126 samples.
Light cache base memory usage is 0.8 MB.
Using premultiplied light cache.
Using adaptive lights ver2
Adaptive light grid built in 7.15 ms.
LightCache contains valid adaptive lights data. Building GPU adaptive lights...
Lightmap successfully loaded.
GPU light cache: 4465 samples stored.
GPU adaptive lights v1 built successfully
Progressive sampling base buffer memory usage: 2016.92 MB
Progressive sampling base buffer memory usage: 2016.92 MB
GPU adaptive lights v2 built successfully
Noise threshold goal lowered to 0.166667
Noise threshold goal lowered to 0.125000
Noise threshold goal lowered to 0.083333
Noise threshold goal lowered to 0.050000
Noise threshold goal lowered to 0.027778
Noise threshold goal lowered to 0.014706
Noise threshold goal set: 0.010000
Number of rays per pixel changed to 103, total number of samples is 40376.
Number of rays per pixel changed to 162, total number of samples is 40014.
Number of rays per pixel changed to 223, total number of samples is 40140.
Number of rays per pixel changed to 348, total number of samples is 40020.
Number of rays per pixel changed to 477, total number of samples is 40068.
Number of rays per pixel changed to 512, total number of samples is 26624.
Number of rays per pixel changed to 64, total number of samples is 0.
Progressive rendering finished: maximum sampling level reached
Light Cache Statistics
Number of premult light cache samples used : 11701991
Number of light cache samples used : 0
Number of failed usages of light cache : 5180728
Closing RTEngine
Max. Mpaths per second: 15.116077
Sampling level reached: 19687 samples per pixel
V-Ray GPU kernel render time: 232.49s
Total GPU samples 2677389156
GPU Memory Stats:
  Device[0] : Tesla K80
    Textures       : 12.40MiB
    Pinned         : 290.00MiB
    Misc           : 1.46MiB
    Internal state : 2820.50MiB
    LightCache     : 0.74MiB
    Total          : 3125.10MiB
  Device[1] : Tesla K80
    Textures       : 12.40MiB
    Pinned         : 290.00MiB
    Misc           : 1.46MiB
    Internal state : 2820.50MiB
    LightCache     : 0.74MiB
    Total          : 3125.10MiB
OCLBufman stats:
Owners count     : 7
With replace ptrs: 6
Geom pointers    : 5
Light pointers   : 2
Textures pointers: 5
NVLink allocated : 0
Misc pointers    : 10
Total pointers   : 24
Device 0 statistics:
  GPU time: 74.162738 %
  GPU ray trace time 80.518431 %
  Transfer to GPU time 0.033748 %
  Transfer from GPU time 19.447821 %
Device 1 statistics:
  GPU time: 70.097189 %
  GPU ray trace time 76.847798 %
  Transfer to GPU time 0.006292 %
  Transfer from GPU time 23.145910 %
Device 0 has 11448MB memory
Device 1 has 11448MB memory
Device 2 has 4096MB memory
VDenoise: Using device 2: GeForce GTX 970
VDenoise: Driver version 461.40
VDenoise: OpenCL compiled successfuly
Render elements to be denoised:
    RGB color
    Alpha
Denoising complete in 44.97 s
Number of raycasts: 4531335 (0.28 per pixel)
  Camera rays: 2250517 (0.14 per pixel)
  Shadow rays: 1134956 (0.07 per pixel)
  GI rays: 1173940 (0.07 per pixel)
  Reflection rays: 0 (0.00 per pixel)
  Refraction rays: 0 (0.00 per pixel)
  Unshaded rays: 0 (0.00 per pixel)
Peak memory for frame 0: 4.93 GiB collected for 22 categories
  Peak memory for "Bitmap": 0 B
  Peak memory for "Bucket image sampler": 0 B
  Peak memory for "Dynamic displaced geometry": 0 B
  Peak memory for "Dynamic geometry": < 1 MiB
  Peak memory for "Dynamic moving displaced geometry": 0 B
  Peak memory for "Dynamic moving geometry": 0 B
  Peak memory for "GI": 1.48 MiB
  Peak memory for "Lights": 0 B
  Peak memory for "Misc.": 1.21 GiB
  Peak memory for "Misc. displaced geometry": 0 B
  Peak memory for "Misc. geometry": < 1 MiB
  Peak memory for "Misc. moving displaced geometry": 0 B
  Peak memory for "Misc. moving geometry": 0 B
  Peak memory for "Progressive image sampler": 2.10 GiB
  Peak memory for "Static displaced geometry": 0 B
  Peak memory for "Static geometry": 1.38 MiB
  Peak memory for "Static hair": 0 B
  Peak memory for "Static moving displaced geometry": 0 B
  Peak memory for "Static moving geometry": 0 B
  Peak memory for "Static moving hair": 0 B
  Peak memory for "Tiled bitmap": 0 B
  Peak memory for "VFB": 1.62 GiB
Number of light evaluations: 1158303 (0.07 per pixel)
Maximum memory usage for resman: 0.00 MB
Number of intersectable primitives: 14
  SD triangles: 12
  MB triangles: 0
  Static primitives: 1
  Moving primitives: 0
  Infinite primitives: 
  Static hair segments: 0
  Moving hair segments: 0
Maximum memory usage for texman: 0.00 MB

Additionally, the progress log does not clear previous work when restarting a new render. Or when opening a new 3dm file.