Doing some quick, photogrammetry pipeline experiments with Zepher.
The objective was to blast through the pipeline, to experiment how quickly can an asset be created using this pipeline.
Photos -> Zephyr (Computing) -> Agisoft Delighter (Delighting) -> Maya (Retopo) -> Marmoset (Baking) -> Substance Painter (Cleanup) -> Engine
Using the tutorial video and the images supplied. (https://youtu.be/17UTelgZqBg)
Threw them through Zepher Free Trial, which allows 50 free images.
Alignment: 5 Minutes
Export and Texture: 5 Minutes.
Dropped the raw scan from Zephr into De-Lighting tool and painted a 20-second mask identifying shadow and lit areas.
The final results for such a quick job turned out great. The delighting tool works wonders compared to other workflows I've used in the past.
Delighting: 4-5 minutes
Here are some more quick breakdowns.
I mocked up a low poly model in Maya, which took about 5 minutes. And baked it it marmoset toolbag.
Finally threw it through. Substance painter to remove the baked grass and other artifacts. (This took another 10 minutes, due to fiddling with some normals from the projected grass).
Cleaned up the model in substance painter ^
Finally, here is the Sketchfab model to check it out. I've chucked it up for free to download. The final texture res is 2k.
The pipeline is unbelievably quick, and affordable considering the other options out there. Zephyr, allowing 50 images free, 8k export is amazing. considering the other free alternative 'Meshroom', which is open-source. It takes double the amount of time process and there is a lot of cleanup involved before you can even consider baking.
The Agisoft delighting tool which is free is a game-changer, as the direct sunlight that was present within these tutorial images would have easily been too much effort for what it was worth. 100% introducing this into my pipeline, however, considering the texture did become a tad bit overexposed in this test so it would be good to explore if there are ways tweaking it.
Substance Painter's clone stamp tool was essential in covering up the projected grass on the stump. However, the normals were the most challenging as I ended up painting over a generic normal intensity over problematic areas. So There's probably room for improvement there.
All n all pretty successful took just under 1.5 hours in total. with some experimentation in-between stages within the pipeline. *coffee etc.