Maximizing the Docker prefs for rendering 118 images

Hello, I’m still relatively nw to using the WebODM interface but have had great success with one of my first projects. Now, as I attempt a 118 image (all jpegs) processing, which includes both nadir and oblique passes, I’ve run into an issue. I received the common Error message when first processing the batch images, which typically is running out of RAM. My laptop has 32GB and is multicore so I went into Docker prefs and hiked the memory to 22GB, leaving something for system memory. It had been at 16GB I believe. I also increased the disk size to 136GB but now, after rebooting everything and restarting, no Task Output log generating, nothing happens. No clock, no green bar in WEBODM indicating processing, just the constant sound of the CPU fan running. I’m wondering if I fully understand what each of these settings is and should be. Could I have overcommitted them? Screen grab attached. While I await some advice, I will lover the settings back to what they were and reduce the number of images I’m trying to process, not something I want to do for a good map and 3-D rendering. I’ll also set the min number features back to 8000 from the 16k I have it set at now. Thanks for any help, and can you also tell me where on a Mac, I find there output log in order to offer it for help? Thanks much all. ==========

==========UPDATE several hours later… I reverted to the previous settings for Docker, and decrease the min number of features to 12000 and in 1:34, it delivered a product. You’ll see from the attached images though that roads and rooflines are distorted or don’t line up properly. Previously when I had this issue, I increased the min # to 16000 and offered more than just nadir shots. For those of you with experience, pls take a look at the task log attached and the screen grabs which highlight the flaws. I’m getting closer for sure. Also, I included the overall mapped result to see if there is any way to not crunch data on such a wide area when I was mapping mostly what is seen in the file “cropped area”.

For distortion of roofs, you could try to increase texturing-nadir-weight.

1 Like

Will do, thanks Piero. The latest 16000 version looked much better but I’ll try the nadir weight suggestion on the next version. I plan to run this data multiple times with different settings.