Cheetah3D 7.2 Beta 3

Martin

Martin

0
Cheetah3D 7.2 Beta 3

Hi,
Cheetah3D 7.2 Beta 3 is mainly a bug fix update. With some luck this will become the final v7.2.:smile:

Thank you very much the bug reports and the feedback. :smile:

New features in Cheetah3D 7.2:
Code: 
-upgraded form QTKit framework to AVFoundation framework
    -supports modern video codecs
      -JPEG
      -H264
      -HEVC (just on High Sierra)
      -ProRes422
      -ProRes4444
    -better image quality
    -multithreaded export
    -much faster
    -hardware accelaration if appropriate hardware is available
-added .gif exporter to Render Manager
-added support for the iMac Pro with up to 18 CPU cores (and 36 threads )
-added support for IES light profiles
-added support for filmic tone mapping to Falcon renderer
-added "Triangulate" property to Boolean creator object
-added "Triangulate" property to Boolean modifier object
-added support for SVG <circle> tag
-added bevel property to box object
-added bevel property to box object
-added bevel property to cylinder object
-added partial cylinder support to cylinder object
-added support for soft shadows to non-physical light sources in Falcon (doesn't use MIS yet)
    -point light
    -spot light
    -distant light
     -sky light sun
-added more steps to the quality slider in the JPEG export options
-JAS files are now exported with a proper UTI
-improved UI of external file link in properties editor
-removed old Cheetah3D 3.2 light sources (.jas loader converts them to Light objects)
-fixed bug in .obj loader
-fixed bug in .obj exporter (caused by objects with identical names)
-fixed bug in smart folder (could forget file path)
-fixed bug in torus object (could create zero area polygons when N1=0 or N2=0)
-fixed bug in movie save panel
-fixed bug with SVG <polygon> tags
-fixed undo bug when using steppers
-dropped support for Mac OS X 10.6 and 10.7


New in v7.2 Beta 3
Code: 
-added IES profiles to area light source
-added various optimisations to IES lights
-added Javascript Document::fileName() function call
-improved UV->PDF tool (better compatibility with Photoshop)
-improved material drag and drop performance
-fixed bug in IES light (too dark in Falcon)
-fixed bug with soft shadows in point light (blured IES profiles)
-fixed drag and drop bug
-fixed bug in cylinder object (partial cylinders didn't work properly when cover/floor was disabled)
-fixed bug in cylinder object (cylinders with 1 height section didn't work properly when cover/floor was disabled)
-fixed some typos
-some small OpenGL optimisations

Attention:
PLEASE BE AWARE THAT THIS IS BETA SOFTWARE SO DON'T USE IT ON IMPORTANT DATA.

Download link of Cheetah3D 7.2b3:
http://www.cheetah3d.com/download/Beta/Cheetah3D7.2b3.dmg

I hope you like the new features in Cheetah3D 7.2.:smile:

Kind regards
Martin
 
I’m seeing an issue, but I’m not quite sure what the cause is. I think I’ve only seen it since running the Cheetah 3D 7.2 beta (beta 1 and beta 3). Video renders appear correctly in the Render Manager window, but when exported as any video format, they look like the image data is wrapping with the wrong width (even though the video file dimensions are correct). Exporting a GIF gives a correct result.

UY92FUR.png


So, I’m seeing this issue with JPEG, H264, HEVC, ProRes422, and ProRes4444 video formats. I am not seeing this issue with exported GIFs. I’m also not seeing it with all Cheetah documents.

It seems possible it’s a macOS bug. I’m having the issue when running on macOS 10.13.2 (17C205). I haven’t done much testing on macOS 10.12, but I don’t think the issue happens with that OS version.

I just created a new document, to try to replicate the issue, and I couldn’t, but I do have a document that always exports the warped movies, if you’re interested (download the renders and Cheetah doc here).

A bit more info: I copied all the objects to a new, fresh document, and the renders save correctly. This issue isn't affecting my work at all — I have a way to get around it. Just thought the info might be interesting and help chase down a bug, if this is a Cheetah issue.
 
Last edited:
Ouch, thanks for the info. That is indeed a problem. It's caused by render resolutions which are no multiple of 4. You've used a render resolution of 510 which is no multiple of 4. 512 or 508 would work.

macOS loves multiples of 4 when it comes to resolution widths or better 16 byte alignments. That has something todo how SSE works (vector width). For most of the APIs every resolution works. You just get a performance hit if you don't use the multiple of 4 resolution width. But the AVFoundation video codecs seem to insist on a multiple of 4 width.:frown:

I've checked the documentation of AVFoundation but I couldn't find any information on video resolution constraints. So maybe it's indeed a bug in macOS. Maybe I just couldn't find that piece of information in the documentation.

Bye
Martin
 
Hi,
the situations is even worser than originally thought. The video needs a resolution which is a multiple of 16.:confused:

Unluckily the only tiny bit of information I could find about that topic is here:

Link

If anybody knows something more about that limitation of AVFoundation please let me know.

Bye
Martin
 
Last edited:
Oval said:
What is
means?
Click to expand...

Most modern Intel CPU have hardware encoder/decoder for popular video codecs like h264. If such an CPU is available AVFoundation will use that hardware accelerator to speed up movie export. If no hardware encoder is available AVFoundation will use a software encoder.

Bye
Martin
 
@marcedwards concerning the movie export problem. I have a fix for that problem. It will be in the next beta. The exported movie will be scaled up to the next possible width which is divisible by 16. That's no nice solution but it seems to be the only possible.

Luckily all popular movie resolutions like HD, FullHD or UHD have a width which is divisible by 16. :redface:

Bye
Martin
 
Thank you for the additional info and research, Martin. That helps avoid the problem.

Martin said:
Luckily all popular movie resolutions like HD, FullHD or UHD have a width which is divisible by 16. :redface:
Click to expand...
Not all iOS devices do though. I wonder if the bug shows up when screen recording a 375pt wide device?
 
Hi Martin, thanks for the quick update, IES lights work great now.

I noticed some strange effect with the new lights softness feature though:
While sun and distant light don't change brightness
the point and the spot light get darker when softened (independent of IES function).

The images below were rendered with the same brightness but different softness values. When the IES gets removed, same effect.

softness.gif
 
marcedwards said:
Not all iOS devices do though. I wonder if the bug shows up when screen recording a 375pt wide device?
Click to expand...

I don't know. The documentation of AVFoundation is really thin. These restrictions usually come from certain optimisations tricks. It's possible that iOS isn't effected due to a different hardware architecture. But that just a guess.

Bye
Martin
 
misoversaturated said:
I noticed some strange effect with the new lights softness feature though:
While sun and distant light don't change brightness
the point and the spot light get darker when softened (independent of IES function).

The images below were rendered with the same brightness but different softness values. When the IES gets removed, same effect.
Click to expand...

The soft shadow effect is achieved by offsetting the light position randomly. If the point light is close to a wall it's possible that some of the point light samples are moved behind the wall and therefore no longer contribute to the lighting in the inside of the box.

For distant lights not the position is perturbed but the light direction. That's why it isn't effected.

Bye
Martin
 
Martin said:
The soft shadow effect is achieved by offsetting the light position randomly. If the point light is close to a wall it's possible that some of the point light samples are moved behind the wall and therefore no longer contribute to the lighting in the inside of the box.
Click to expand...

Thanks Martin for that explanation.
Indeed lights more distant from surfaces don't show any darkening.

Now most lamps are attached to walls or ceilings and so it's great to have an IES option for the area light too, because that can be configured as a disc, put close to the ceiling and will look right as if built in.
 
You must log in or register to reply here.
Back
Top