Render in Place w/ preserved alpha

Is there a way of using 'render in place' on a fusion clip so that it preserves the alpha?

Right now, I've got a long clip that I'm placing fusion comps on top of that I don't want as part of the comp, but if I 'render in place', I loose the alpha and don't see the underlying clip. Obviously I could render the fusion clip out using a format that supports alpha from the deliver page, but then I lose the ability to 'decompose to original', right?

---

Surface Studio, 4 core i7-6820HQ @ 2.7GHz, 32GB RAM, 1TB SSD, GeForce GTX 980M w/ 4GB vRAM
Dell Precision 5820, 10 core i9(?) @ (?), 64GB RAM, 1TB SSD, 2x RTX 1080Ti SC2 Hybrid - PSU went boom!

It works for me

Oh, cool. What format are you rendering to?

Even I want to know that

Sent from my ONEPLUS A6000 using Tapatalk

DNxHR 444 and ProRes 4444 both work fine and are commonly recommended.

But surprisingly, DNxHR HQ and DNxHR SQ both also work, even though these are not variants normally listed as having an alpha. (Doubly weird as DNxHR HQX - the format above HQ and below 444 in bitrate/quality does not store an Alpha through Render In Place.)

Here's the list of codecs I've tested that preserved a Fusion alpha in Render in Place:

In Quicktime:
- ProRes 4444 (not available on Windows)
- DNxHR 444
- DNxHR HQ
- DNxHR SQ
- GoPro CineForm RGB-16bit
- GrassValley HQ
- Uncompressed ARGB and BGRA

In MXF
- All the DNxHR formats listed under Quicktime

In EXR (image sequence)
- All

In DPX (image sequence):
- Alpha 8-bit
- RGBA 8-bits
- RGB 16-bit
- RGB half-float 16bit

I didn't test every codec and every format so there might be more. H264 and H265 definitely will not support an alpha.

Thanks for the huge info brother but can you also tell me which one renders faster and takes less space?

Sent from my ONEPLUS A6000 using Tapatalk

Thanks for the huge info brother but can you also tell me which one renders faster and takes less space?

All of those formats should render pretty quickly, and all will use a lot of disk space compared to H264/H265.

Of the ones I listed, I imagine DNxHR SQ would use the least space. Though I'm not 100% sure how GrassValley compares.

What resolution and frame rate are you working with? We can work out the space requirements for DNxHR SQ from that.

I just did a quick test at 1080p 30 FPS, and a 56 second clip with alpha, where about 50% of the clip was cut out in Fusion (ie RGBA values = 0), had the following file sizes:

- DNxHR SQ = 1.4GB
- GrassValley HQ 1080p = 833MB

GrassValley also ran slightly faster - 220 FPS instead of 200 FPS on my system

First, let me list down the pc specs and the work, for which I use fusion comps.

PC specs :
Processor : AMD Ryzen 3900X 12 Core CPU
RAM 32 GB clocked at 3200mhz
Graphic Card GTX 1660 6GB V RAM
MSI B550 A-pro motherboard
WD SSD 500GB
Secondary SSD Kingston a480. 480 GB SSD

Why I use fusion?

I make videos which requires a lot of text with animation because of being an explanatory video.

So, sometimes, I have to insert 7 text plus fusion comps one above another in edit page, and this intensive task makes video playback at 2fps and makes me impossible to work

Sent from my ONEPLUS A6000 using Tapatalk

@TheBloke, thank you again for both helpful notes and taking the extra time to explore the boundaries of the software behavior - helping other people not have to repeat the effort.

That worked, thanks all!