I have a production that is 64 gigapixels and made from 1500 42mp photos. Once completed it is very laggy to use. As you navigate in CC Viewer the tiles that you are viewing are slow to resolve and all of the tiles out of the field of view loose their resolution so when you navigate back to them they have to resolve all over again. I made this project web ready and put it on my website. When opening in in a browser it takes about 10 minutes to load and then has the same issues. I have opened models of entire cities and not had these problems.
I asked Bentley support about this and they suggested re-running the production with the 'LOD across tiles' tool enabled. I did this and it was much better, but still a little slow to resolve. They also suggested trying to run the production with the tiling set to 25GB of RAM / tile instead of the recommended 16GB. I am running that now to see if it helps. I'll know in a few days when it finishes up.
In the grand scheme of things a 64GP project wouldn't be considered large in the Bentley environment, so I'm wondering if anyone has run into issues like I'm having and what they have done to correct them. Are there other settings that you use for bigger productions to make them more snappy?
Which Contextcapture version you are using? Viewer version? Web ready means it is 3MX format? If so it is heavy format for filesystem as it contains lot of small files for each tile and need good HDD which has low seek times. If it is SSD then this shouldn't be a problem. Also I would have recommended to clamp resolution as not to have too many details. But sounds as issue with Viewer as the format is meant for unlimited resolution.
Thanks for the reply Oto. I am on the latest version of both CC and CCV. Yes, it is a 3MX. My drive is an M.2 SSD with read/write speeds in the 3300mbps range. My computer has a benchmark around 380. The issue happens on all machines from my production machine to my MacBook Air. I have the same issues working within ContextCapture Master as well as in the Viewer.
Your comment about 'clamping the resolution' was interesting. I haven't heard that term before. How do you clamp the resolution other than running a subsampled production? I have done that and run 50% subsampling. The model works fine at that resolution and is actually pretty good. However, this is an archaeological site where the purpose is to document ancient plasters, pictographs etc, so high resolution was the point. Many of the 42mp images were taken within 3 meters of the subject and when run at 100% resolution the production has amazing detail. The image below is a screen shot of the model, not one of the images. It covers about .5 x .5 meters.
Obviously, when I'm looking at models of whole cities the number of gigapixels would be much higher than what I'm working with, but the resolution of small details would mostly be much lower. I guess now I'm wondering if the number of gigapixels is the determining factor in how a model loads or does fine resolution of small details also play a factor?
My production with the LOD set to be across tiles and the tiles set to 23GB of RAM is about half complete and it is much more responsive than the first production that didn't have the LOD set across tiles and had the tiles set to the recommended 16GB of RAM. So far none of my tiles has failed.
Resolution limit(Clamp resolution) is setting under reconstruction. So if detail of 1mm is enough then do not produce details below 1mm.
But seems less tiles are better. Actually the 16GB limit is only if you need to use retouch tools. If no retouch is needed or desired then you can safely increase to 32GB if you have enough RAM. We have computer with 128GB but tiles larger than 64GB mostly got stuck in production so the 128GB RAM is wasted as couldn't utilize it.
That's helpful. Thanks