MeshTree

got done with the readme and pushed the new dmg. new devlog. about 6 hours left till stardance ship closes, time to ship 🤞

finally done with the chunk reconstruction system and the ram usage is reduced. since it’s kind of new territory for me, documentation was hard to find so AI helped with this thing. here’s how it works in short, it measures ram every 0.3 seconds and keeps a report of it, looks at the quality preset and determines how many chunks are needed, builds them one by one, then merges them into one model. The photoset I am using to benchmark how much memory is reduced showed from a 3 gb ram usage to about 2 gb on average. I also tested with multiple reconstructions to make sure it actually works and nothing breaks.

Working on the ChunkReconstruction system for better memory management

got rejected for the 3rd time😭 reshipping again🤞

Added a remove button to remove individual pictures. Fixed some bugs and polished a bit, I’m using the same photoset each time as a benchmark ik it looks a bit odd😭. Now shipping for the first time 🤞

Been a long time since my last devlog, was busy with exams. I realised the blur detection wasn’t working properly, soo I’ll have to work on that. I also noticed it’s using a massive amount of memory for a relatively small render, so most of the time went to just testing how the blur detection actually detects a blur and the memory load, and watching activity monitor while rendering, so progress might seem slow. But I have plans to make the memory situation better by using a chunk system where only a small section loads and renders soo there’s less load on memory. Also, since I tested a bunch of renders from objects around my house and our car, taking the pictures was a massive pain, so I’ll work on MeshCam so I can take pictures automatically at a set interval. I also have plans to add video support tho I haven’t figured out anything for that.

Been a while since my last devlog, i’ve been busy with other things, but still managed to give some time to MeshTree. Last time I got Object Capture working with a very basic UI, just to see if the pipeline works or not, thankfully it did. Since then i’ve added a ton of features :
-Quality presets(Low,Medium,High,Max,Ultra)
-Cancel button
-Blurry/low res image detection
-3D model viewer
-Progress Bar
-.usdz export
-A bunch of UI tweaks (this one alone took a surprising amount of time)

Up next:MeshCam:-
I’m planning a companion app called MeshCam. The idea is it will capture photos at a user selected interval and log in motion data, all in a single file that the user can just airdrop to their mac and MeshTree, and use the motion sensor data for even higher quality models

Object Capture is working, successfully generated the first local .usdz 3D model from an image folder. Had some issues with WakaTime logging time, but after a bit of troubleshooting, it’s now fixed.