Nested Compound Clips, Drop Warps, & Rendering, Oh My!

I am sharing this as a lessons learned for other people so you do not find yourself in rendering hell as I have the past couple of months.

BOTTOM LINE UP FRONT (BLUF): Obscenely slow rendering times may be the result of bad timeline design decisions on your part and not your hardware or software. Pick the right tools from your toolbelt to avoid this painful lesson learned from me!

Over the past couple of months I have been finding the rendering of a 1:30 to 2:00 minute video taking upwards of 16 minutes, with one 10 second clip taking 5-6 minutes, and thought it had to be my hardware and/or software configurations when I upgraded to DR Studio 19 and Windows 11 at the same time. In no way could it have been something I was doing because the use of a fusion driven transitions (Drop warp) was minimal. The amount of time I spent troubleshooting the GPU and NVIDIA drovers I will never get back.

Oh but I should have remembered an adage from the days when I provided user support for software applications: Users can and will find new and exciting ways to fudge up an application. And sure enough it was my kludge approach to this full page graphic:

A Graphic clip built in Davinci Resolve that gave me rendering hell

fullpagegraphic.png (152.99 KiB) Viewed 501 times

And further rendering slowness with a ten second clip that had these two full page graphics:

Another Graphic clip built in Davinci Resolve that gave me rendering hell

DRGRaphic_Render2.png (149.55 KiB) Viewed 501 times

Another Graphic clip built in Davinci Resolve that gave me rendering hell

DRGRaphic_Render3.png (308.01 KiB) Viewed 501 times

For each of these graphics I was using the Drop Warp Transition.

So what was I doing wrong?

Yesterday I went to see what I may have done to create this rendering mess.

In the first example above, I had it using nested compound clips. I did this to maintain my sanity building them because DR really was not the right tool to use to build it. I wanted to compartmentalize each section as much as I could and crazily though that I could use the compound clips as Pseudo-layers. This turned out to be a bad decision for reasons beyond rendering, but that is a topic fur another day also tied into Pseudo-layering.

I decomposed them all (well almost all) in place. For some reason one would not decompose so I had to open it in its on timeline and decompose there. I then copied them all over to the main timeline.

It was over 80 tracks!!!!

I thought, "did I really do that to myself?"

I sure did and apparently it takes the DR rendering engine time to apply the drop warp transition that I applied to the base compound clip to everything nested many layers deep in the nested compound clips. It was not that it was taxing the GPU (I checked that real time), it was just that it was a time consuming process!

The same problem existed for the 2nd and 3rd images above for the same reasons!

These graphics will not be used again until next year, and i wondered what options I might look at next time I have this situation and how would they test out.

Option 1: rebuild and Manage the graphic in Photoshop or Gimp. Dumping out the image and replacing my compound clip with this exported PNG took the rendering down to a few seconds.

Option 2: Rebuild the graphic in a real Character Generator designed for things like this and dump it out as an AVI, use the AVI instead of the compound clip. I am using Ross xPression as a CG for other projects. I took a prebuilt complex sports graphic from an existing project that included video, audio, and still elements, and dumped it out as an AVI file. I imported the AVI file and replaced the compound clip with it. Once again, it rendered in a few seconds.

And I am sure there are other options and combinations I can do, but my bottom line advice:

Do not try to build and use complex graphics like this natively in Resolve. Especially the way I did it. It will not end well and you will be cursing at the rendering times and asking yourself "what rendering gods have I offended!?"

Use external tools better suited for graphics like these and let Resolve do what it does best!

Hope this is helpful for people out there.

While its admirable you shared a lesson, something not often found in these parts. So kudos for that. Your comment about "Do not try to build and use complex graphics like this natively in Resolve. Especially the way I did it. It will not end well and you will be cursing at the rendering times and asking yourself "what rendering gods have I offended!?""

I would say that much much more complex graphics can be built and rendered and not break the bank or machine if you use proper workflows and methods, so there is more to learn. I don't see anything really problematic in your screenshots that resolve / fusion could not handle. Unless of course one uses brute force. But for sure, I fully agree that when users use these tools as if there is no differnt in what comes before what and in which manner, things become a serious mess with slow rendering times. But virtually all of it can be optimized and made run much faster than most people realize, if done correctly. Once again, kudos for sharing a story about lessons learned. We need more of those.

BOTTOM LINE UP FRONT: Obscenely slow rendering times may be the result of bad timeline design decisions on your part and not your hardware or software. Pick the right tools from your toolbelt to avoid this painful lesson learned from me!

That's true. There are a lot of potential workflow issues that may ensue if you choose a bad path. It helps to keep things simple and try to avoid all the potential roadblocks up front. Particularly in the case of Fusion graphics and composites, we try to render all those separately, then bring the flattened files back into the Edit timeline so there's no slowdown and no potential Media Management issues.

Nested Timelines are another potential landmine, and we try to avoid those as well. For us, we generally try to just render out a flattened mezzanine file in high-res 444 (or 444XQ), stitch all those together, and then deliver the completed long-form project -- episodic or feature film. I wouldn't try to stack 5-6-7 nested timelines and expect them to work flawlessly. It gets even more dicey when sound is involved: we try to keep all audio original and unrendered, so the continuous WAV file(s) is dropped in at the bottom of the Edit timeline with all the pre-rendered picture files on top.