My overall goal was to build a repertoire of reality capture tools and processes. I have greatly enjoyed discovering new processes and am wrapping my brain around the realistic scope of something like my mock project last term.
Testing scanning hardware
- For both scan quality and hardware accessibility reasons, I have concluded that photogrammetry with digital cameras are the way to go for my work.
- My Google Pixel 2XL has a decent camera with 12 megapixels
- I prefer to use the ACELab's SonyRX10 camera with 20 megapixels
- I used the Ricoh Theta for 360 photography
Scanning software
- I am using 3DF Zephyr in combination with Maya, ZBrush, and Substance Painter
Researching pointcloud rendering
- It was suggested I look into point based realtime rendering to render the pointclouds directly (instead of optimizing them into polygon meshes). I discovered a couple of tools/plugins in development, but at the moment none look like they would be useful to me in the short term.
- The concept is very exciting though and the potential for a more scaleable resolution in the models would be amazing.
- I found a scientific paper that outlines the whole process of macro photography and focus stacking. They even published their tracking ring.
- I wanted to scan another bug, but I put this away for now as I did not feel it was directly relevant to my larger term goal.
- I would be interested in experimenting with lightfield photography for this application.
- This is the easiest and most common application of photogrammetry.
- My most successful capture was a statue in the woodlands cemetery which I named the "Lichen Angel" since she was covered in lichens.
- For regular camera I used a combination of shooting patterns, starting with three rings outwards, then one ring up/down. Then detail passes in simple paths.
- For 360 camera I found up/down rings are better.
- I'm looking for enclosed or semi enclosed spaces with natural elements, interesting forms and texture.
- Shot a few locations:
- Main Building's smoker deck (successful)
- Travis's new apartment (unsuccessful, bathroom was too shiny)
- Woodlands cemetery (somewhat successful)
- UPenn Courtyard (somewhat successful)
- Scouted potential locations:
- Main Building's observatory -- need special permission to access
- UPenn's Bio Pond -- there are few enclosed spaces and PLENTY of natural elements
- I have gotten very fast at shooting bust portraits. I posted a bunch on my Instagram.
- I have the subject sit in a chair in a well lit room. I stick tracking stickers to the back of their head, as most hair creates stitching issues. I shoot in continuous/photoburst mode and walk around in three rings.
- I would like to take this idea further, either cleaning scans or at least hosting the pointcloud data somewhere like Sketchfab
- I enjoy the ritual of this process a lot and interacting with friends who will sit for me.
Misc
- Found an fascination with technopaganism, its somewhat silly and quotable, but I feel like its a staple of 90s cyberpunk subculture. Would love to find more on this topic. Inspiring for interactive design.
- Wrote about the walking simulator genre, and why I love spatial design
- Found a growing field in volumetric and spatial capture
- the future will be lightfield photography, giant capture domes
- volumetric video capture is exciting
Next steps
- Could revisit/refine really anything above.
- Need to push the Smoker Deck room forward.
- Retopo
- Bake maps to low poly mesh
- delight albedo
- adjust materials
- flex goal - object seperations
- flex goal - interactive elements
- Would like to work with more interactive design this term, however I think I am running out of time.
- The data capture side is the fun part! I need to push a model through to completion.
- It may make more sense to revisit the Lichen Angel capture for the post processing steps before tackling the whole room.
No comments:
Post a Comment