Hello, 

Can you draw a small diagram that explains the workflow some more? 

Hi,

You can use the alpha splitter module on the output.

Best

Hi,

As Eric wrote if you have an image with alpha you can split it and output both channel individually.

Unfortunately however the output of the TrackedCam_AR_Prev_3-Cam already contains the "background" which is the input camera image.

We can try to create a version of this camera that doesn't have a background, but it is not clear how would you use it.

Do I understand correctly that you wouldn't have a physical camera (attached to Aximmetry) in this case? 

Also would you use some hardware position/orientation tracking?

Would you render 3d elements with Aximmetry or just 2D overlay?

Best Regards

Hi, 

I tried an alpha splitter module: with 2D content (with alpha) it works well, with unreal scenes I couldn't achive the same result.

We can try to create a version of this camera that doesn't have a background, but it is not clear how would you use it.

• The advantage of the TrackedCam_AR_Prev_3-Cam with Alpha  would be that I don't have to process the original camera image with aximmetry.

Do I understand correctly that you wouldn't have a physical camera (attached to Aximmetry) in this case? 

• The AR-foreground and the original camera image a background are mixed together in the production switcher. Camera tracking will be used (for the initial setup, the camera image directly in aximmetry helps at this point). 

Would you render 3d elements with Aximmetry or just 2D overlay?

• We want to use both 2D content and 3D unreal elements as overlay.

Thanks for your support, 

Best

Hi,

We created a prototype that might fits your needs. Can you please check it?

https://mab.to/44Bak62Wf

A few notes:

- There are two new outputs: Color and Alpha. You should output these two in two different (SDI) output.
- Reflection and shadow of the AR objects will be only an estimation, because it is physically impossible to do them correctly as an overlay.
- Our output has SRGB transfer function so your device that will do the blending should convert it to linear when the blending happens to have a correct result.
- The color image is not premultiplied (by the alpha). I think this will be the correct one for you, but if needed you can add a few more modules  to make is premultiplied.

Best Regards

Hi, is there a solution for this available now? Output Fill and Key on separate SDI Channels without the background image of the tracked camera? (Augmented Reality) Thx and best regards. 

Hi,

You can now actually do this without modifying the camera compound.

However, you still need to do some tricks. First, you have to blank out the video input. You can do this by copy-pasting the blank value from the Zoom Device parameter:

Into the Camera Device parameter:

Then disconnect reflection's that you don't use. If you don't use any reflection catcher in Unreal, then disconnect all three.
And use the Alpha Splitter module with the Unpremul pin turned off:

The notes from the previous Aximmetry comment apply to this solution too.

Warmest regards,