Render fails depending on the camera position

[joshua_symonds1]

If I run a render from a distance, Vray has no trouble at all. However, if I move the camera closer to the object, all of a sudden Vray throws up some error messages and fails the render. Literally not changing anything other than the position of the camera.
I'm still fairly new to this software so forgive me if I am missing something obvious, but I don't really understand how changing the position of the camera would have any effect of the performance of the render engine. I have attached a screen grab of the error log. Any help would be greatly appreciated.
Thank you!

[tsvetomira.girginova]

Dear joshua_symonds1, hello.

"Depends on the viewpoint" is not only wisdom.
Changing the camera position is changing the task which should be calculated for rendering.
From your explanation I could guess that you may using V-Ray Fur or V-Ray Displacement, but let's stick to the specifics.

The screenshot from the V-Ray Log is showing error that your scene is consuming memory that is not available on your configuration.
It is shown that you are using V-Ray GPU rendering. Here advices are shared how to optimize your scene.

In order to be more helpful please provide us the following information:

• exact host platform version
• exact V-Ray version (screenshot from Extensions > V-Ray > Help > About)
• are you using CUDA or RTX rendering
• graphics card type and currently installed driver version

We could take a look at your scene if you share it with us.
Please use Extensions > V-Ray > Pack Project to collect all the assets and create an archive.
Open a ticket in our system from here and choose 3D and rendering help under Request type.
Mention this thread in the description.

Waiting for your answer.

Best regards,
TS

[joshua_symonds1]

Hi there,

Please see the below answers to your questions:

Exact Host platform - Sketchup Pro 2021
Exact V-Ray version - Latest update of V-Ray 5
CUDA or RTX - I have tested with both CUDA and RTX with the same results
Graphics card type - Nvidia RTX 3070 laptop GPU 8GB VRAM on the latest game ready driver

I have attached screenshots of my Sketchup, V-Ray and GPU info.

Would you be able to elaborate on how adjusting the camera position effects the calculations for rendering as I can't really wrap my head around it. The resolution is the same for both the render output and the textures themselves, I don't really understand how moving the camera can have that much of an impact? If anything, moving the camera closer to the object should be better as there is less in shot for V-Ray to render? or does it render everything regardless of whether it is in shot or not? Either way, I'm not fully understanding why the camera position makes a difference.

I will pack my project and upload it to your link.

Thank you for your help.

[joshua_symonds1]

I have now sent over the packed project file, ticket No. 36359. Thank you!

[tsvetomira.girginova]

Thank you, joshua_symonds1.

I am looking into your file now and could share the following:

• be informed about the recommended GPU driver version from this webpage - now 497.09 is the one
• the log is showing a message related to vRAM usage - insufficient memory
• as supposed initially, the difference in vRAM usage is due to Displacement - in your case used in materials (Wood Siding, Veranda Retaining Wall, Sand, Roofing Tiles). Important to mention is that these materials are used for large surfaces. So, the Displacement impact is the main cause of this thread.
• it is easy and helpful to check the Stats section in VFB's right wing, like this, where you are able to see the Used memory

Conclusion - the scene should be optimized by using some of the approaches described here & here.

Main advices are:

1. Check the difference in vRAM usage when Displacement is globally turned off from here
2. Use On-demand mipmapping instead of Full-size textures
3. For this high-resolution image (4K) and currently available vRAM uncheck View Dependent from the Displacement settings of the materials
4. Take a look at the i sign before the Displacement, see here

Please take into account these advices and perform more tests to see the difference.
Hope the situation is better explained now.

Best regards,
TS

[joshua_symonds1]

Hi there,

I taken some time to go through your advice and run some tests. Using On-demand mipmapping I am able to reduce my vRAM usage enough to produce renders more consistently.
I hadn't realised that I could use a geometry displacement modifier instead of the regular diaplacement map, what are the differences between the two? other than how they are implemented. Does the modifier have better performance?

Also, I wonder if you could explain what the view dependant feature actually does, as I am not able to find any information online.

Thank you for your help.

[tsvetomira.girginova]

Hi again!

Thank you for sharing. Happy to now that you saw the difference and you are now aware what could be done in similar cases.
As visible, the attribute is a legacy one. In some cases and different platforms exporting this kind of shaders isn't the best thing, so something new is on.
Using Displacement as Geometry is the improved and recommended way. It gives you also the freedom to set it up separated from the material and apply to more individual objects.

About View Dependent - it is related to Edge Length and was mentioned in your case especially for the big resolution of the image (4K).
You could check the information here & here.

These were quick advices, but I think the links above will be helpful to you.

Best regards,
TS