Krk Croatia - apartment 2

196

[X5, 72MP, 8K30FPS) FRAMES: Scene created from three passes (low, mid, high) supplemented with an additional detail pass for spots where I wanted more detail (such as the shelf in the living room or paintings). Each pass was about 4 minutes, average 270 extracted frames, detail pass had 539 extracted frames across 17 short videos. Furthermore, I used 125 HDR stills shot on a different day to improve the splat. I used only stills that had similar enough exposure, and Spirula accounted for the remaining difference. So, 2831 frames total, 2706 fisheye images (1353 dual-lens frames from four video passes: low, mid, high and detail) + 125 HDR 72 MP 360° stills. Reconstruction registered 1349 of the 1353 video frames per lens and 121 of the 125 stills. SETTINGS: Made this with calculated normals and geometry settings, but before I learned about --distraction-robustness so I still hand-masked all the shadows (which is a pain in the ass, long live --distraction-robustness). Changed these settings between 2-3 runs because the geometry would come out weird, floor folded into itself or floaters made into solid bulges in walls where lighting was lower, thin objects like shower door or nightlamp disappearing, floor tiling edges collapsed into one flat surface. There's artifacts like that in this splat but I haven't had the chance to rerun the splat with different settings yet, so I'll replace the splat when I do that. PROBLEM SPOT: Biggest challenge is the hallway part with the mirror and flowers. I set manual exposure settings according to the living room as the brightest spot, so understandably the hallway is a lot darker, even with the lights turned on. I assume white balance difference also plays a role. That's why training struggled in those spots where it's just a blank wall and darker corner. Curious to hear from you, how would you fix that? Would you segment the apartment into parts that call for their own exposure settings and then shoot separate videos using these settings, one video per height? The obvious thing is use lighting positioned in dark corners to mitigate this, but I don't have the budget for that now.

Software Used

Spirula Studio

Suggested

1 Comment

takav1 month ago

Posted as comment because of character limit in description REFLECTIONS: Furthermore, I used AESUB orange spray on the mirrors to kill reflections. It worked great, but I'm not sure how feasible it is unless you're working at scale since one can cost me €45 and I'm not sure how long it will last me. Also, there's the question of the potential client being iffy about this stuff applied to their mirrors/glass because of damage. There was no damage to the mirror or frame, but these people don't know ish and might be squeamish about it. I might be overconcerned, but I did this shoot in an apartment from my friends parents we vacationed in. When she found out they're stopping by she got all riled up about removing the spray from the mirrors so they wouldn't see it lest they might be shocked about what I did and suffer a panic attack because of the ruination of their mirrors. So I went to manually remove the spray and actually learned a few things (thanks boomer parents): If using AESUB, absolute best course of action - do nothing. The spray really sublimates on its own completely. Orange takes the longest, maybe a day, so that might concern someone, but I tried leaving it on a spot at home and it went away on its own. I didn't know that when I was at this apartment, so like a chump I went on to force remove it and let me tell you, what a pain that was. I used a damp cloth to scrub it away, and it mostly worked. But there were these stubborn "hard" spots left that just wouldn't go. I had to scrape them off the mirror with a plastic spatula, wiping with the cloth as I go. And it took forever. I didn't wanna risk using dishwashing soap or similar because I didn't know how they would react to each other, and I didn't have any 70% alcohol which I assumed would work like a charm (tried at home, it did). I assume that the hard spots are the result of repeated application of AESUB on same surface, so the slight left-overs from previous shooting were caked over with the new layer, cause these layers seemed like they were baked into mirror's surface. So there you have it, learn from my mistakes.

Loading comments...