temp folder size in function of GB pictures?

I have about 150 GB of pictures (18000). What needs the size of the temp folder be to process these pictures on high detail?

Good question.
Is your capacity limited?
Because it probably won’t be more than the 150 GB of source images, if you only do one calculation…

Hi Tom Traen

have about 150 GB of pictures (18000). What needs the size of the temp folder be to process these pictures on high detail?

are that JPGs of TIF files or ?

The pictures are in .jpg format. When I want to process 7000 pictures I have reach my limit of 250 GB and reality capture ran out of memory…

Hmm. I was way off then… :oops:
But how full was your cache when you started?
I know I can work quite a while with 300 GB, but then I never had more than 2500 images in one set.
Even then it took a while to fill it up.

But honestly, if you are looking at such datasets, why don’t you just tuck a dedicated cache drive in there - at least 1, better 2 TB?

i find myself at 650 - 700Gb all the time.

I’ve had a few time go past 1.2tb

but I find if i reconstruct one camera set, in multiple regions, it creeps up quite as lot, as I can’t clear it between reconstructions since i need to keep depth maps.

Hi Tom Traen

was hoping you post even image resolution as that is most crucial to say what is UNCOMPRESSED data volume and what to expect as JPGs are compressed so even harder to say some numbers.

But be prepared for 1 TB of free space for HIGH recon mode.

Hello,

 

I have component with 6650 images (jpg), I’m doing reconstruction on normal detail (image downscale for calculating depth maps is at 1). At the moment, process of calculating depth maps is at 77%, and temp folder is 500gb. Does that mean that at fully calculated depth maps, temp folder will be at roughly 670gb, or do I need extra space (how much???) for something else? Temp folder is at ssd partition 780gb :frowning:

Is it possible to add second (backup) temp location, so we don’t lose hours (in my case days) of processing time?

Best Regards!