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.

Topics - Njen

Pages: [1] 2 3 ... 5
Gallery / CYAN EYED - Animated Short Film
« on: 2021-08-21, 23:06:29 »
It has been a long time coming! CYAN EYED is finally available online here:

Over seven years in the making with a crew of over 100 people including the geniuses from Skywalker Sound, I am happy to finally have everyone watch it! Everything except for the background environment (landscape and sky) is rendered in Corona

Also some good news as the film has just started its festival run, yesterday it won Best Short Film in the Australian Effects and Animation Festival Awards (an international award competition):

Please watch, like and share, thanks!

Official website:

Gallery / Cyan Eyed Offiical Poster
« on: 2021-06-09, 22:06:23 »
Hey everyone, today I released the official poster for my upcoming animated short film, Cyan Eyed!

And for those that might have missed the trailer to the film, here it is:


I have a scene that has moving geo, and I have animated the camera to move along with it, so the relative motion blur is almost non-existent, which is correct.

I have a vdb smoke sim that is simmed in world space, so the vdb grid is not linked to anything, and also tracks with the objects and camera. But when I render, the vdb appears to have motion blur that is not relative to the camera (first attached image).

To confirm this, I deleted all keys on the camera, and did another render. Now all of the geo is blurred because the camera isn't moving any more (which is correct), but the vdb has no motion blur, but it should, as it is moving through world space at the same speed as the geo (incorrect), which is basically the opposite of the previous image (second attached image).

I have set up a caustics pass render for just the eyes of a character. Everything in the setup works fine except for what I can only describe as random buckets that go missing every time I render the image even though nothing changes. As in, I hit render, and one bucket will be missing. Then I hit render again, and another bucket will be missing.

A description of the attached images:
  • 1_viewport.png - The geo that is included in the render.
  • 2_beautyPass.png - The beauty pass. No caustics are enabled in the beauty pass, just the caustic pass. This image is included just for reference.
  • 3_causticAOV.png - The caustic pass. In the top right corner a 'bucket' seems to be missing.
  • 4_causticAOVExpose.png - The caustic pass exposed up. Checking to see if any data does actually exist in the missing 'bucket'.
  • 5_causticAOVRerender.png - Nothing in the scene was altered in any way. Rendered the image again after the previous one finished, and a different 'bucket' is now missing information in the lower right area.

I have come across another issue with VDB's, where the "Visible to Camera" attribute doesn't seem to be respected. In the attached image, quality issues aside, on the left the rocket smoke VDB "Visible to Camera" attribute is turned off, and on the right it is turned on.

Regardless if the attribute is turned off or on, I would expect the VDB to cast shadows on to the surfaces, but it seems that with the attribute off, the VDB seems to not even cast secondary rays as well as primary rays. More specifically, I would expect the surface area 'A' to have a similar amount of shadow from the VDB as surface area 'B'.

It seems that some AOV's can not be used properly when a VDB grid object is in the scene. If there is pixel coverage of a VDB object, I would expect that coverage to hold out what ever data is behind it. As it is currently, I can not use a number of AOV's in comp because it is not being evaluated in those AOV's.

For example, I can't use an RGB mask to isolate other objects behind the VDB grid. Other AOV's I would expect the VDB grid to hold out is in source elements and texmap elements. There are more as well, though I'm not sure how they would be rendered correctly (zdepth, UVW, position, etc.).

In the attached example, there is a VDB grid of rocket smoke in the middle of the image in the beauty pass, but there is no evidence of it in the attached AOV's.

Just checking if this is currently expected behaviour, or if it's a bug that could be fixed. And if it is currently expected behaviour, could this be added to the list of fixes for the next version.

If I have a VDB grid that has an emissive portion, then set that grid to not be visible to camera, then the grid doesn't cast any light on other objects in the scene. I would assume like any other self illuminated object, light or object light material, then it should still be able to cast light when not visible to camera.

I installed Corona 5, and I've seen a what I call 'RAM Creep' when network rendering (using Deadline), where each consecutive task that a slave renders sees a little increase in RAM until it hits maximum and then the machine starts to use swap. I requeue the slaves manually, and the RAM goes back to normal for the first task, then starts to creep up again. In the attached image of one of my Deadline jobs,  I have three slaves, and all of them saw increases in RAM (last column) until I manually requeued them, and the same thing happened again.

I'd be happy to upload a scene if you need it.

[Max] Bug Reporting / CVolumeGrid visibility track...
« on: 2019-09-29, 11:33:26 »
It seems that the visibility track on CVolumeGrid objects is being ignored. I animated the visibility attribute, but when I render, these objects are always visible, no matter what value the visibility attribute is.

Can this please be fixed?

So I've found three different ways to derive the vertical FOV (field of view) from the horizontal FOV, and each method gives me a slightly different number. So what I want to know, is which one is the correct method?
Code: [Select]
vFov1 = 2.0 * atan(tan($cam_render.fov / 2.0) / GetRendImageAspect())
vFov2 = atan(tan($cam_render.fov) * ((renderHeight as float) / renderWidth))
vFov3 = ($cam_render.fov / GetRendImageAspect())
So, if I have a 70mm camera, using a 36mm film back, and a resolution of '1920x803', the resulting values for each of the above are:
Code: [Select]
vFov1 = 12.2765
vFov2 = 12.9701
vFov3 = 12.0624

I have scenes that crash after two or three rerenders (Using the 'Render Last' button). I'm pretty sure it's related to a character asset that has Ornatrix hair. It crashes during the parsing stage (Max 2019).

I am uploading the minidump file to the private uploader. The file is called:


One handy little feature that Vray has with Deadline is that the little progress bar for a Task shows the percentage of the image complete, as well as the bar itself that changes from green to blue based on the percentage. With Corona, it's an all or nothing UI feedback, where all tasks in progress show 0% complete until the frame finishes. It would be a nice update to the UI if the progress bar for the task shows dynamic information when using Corona with Deadline.


I have a scene that renders fine when rendered through the Max GUI, but when I submit the scene to render through Deadline in slave mode, it will randomly crash a lot at the parsing stage. I'm pretty sure it is some object, because when I hide groups of objects, sometimes it renders fine. But because I have thousands of objects, it would take me a while to constantly send scenes through Deadline to find the offending objects(s). Is there some way to increase the verbosity of the output during the parsing stage?

Is there any reason why the 'Radius Scale' on the skin shader isn't mappable? If not, can I make a request that it can be?

The R, G and B isolation channel buttons in the Standard Max VFB, and Vray's VFB are super handy. Can we please have them in Corona?

Pages: [1] 2 3 ... 5