From this link, I think Epic would not be happy with such a pull request.
Is performance indeed better when using stationary lights than your animated lightmaps technique? (I was aiming for mobiles, but after reading Epic’s response following your link I am not sure what to do now)
For mobile the major challenge for MLDB is: the storage size for lightmaps. When the number of lightmap sets goes up, the size of umap file may explode.
For rendering performance, MLDB will be definitely slower than standard lights. However, it’s a ‘quasi-dynamic’ GI solution, and from my own point of view it may be the only ‘dynamic’ GI solution available on mobile platforms.
Either way, having multiple lightmap sets would increase your file size by a lot, which would probably be important for mobile.
As long as all of those lightmaps are not stored in memory at once, it shouldn’t be an issue.
I wonder why not to have 1 neutral set of HQ lightmaps, and several LQ blending lightmaps just to simulate GI ?
Well, it has to store some of it in memory at the same time, if it unloads a lightmap before it loads the new one then it won’t have anything to display, but it could do it in a way so that it’s not doing all of the lightmaps at once
You wouldn’t be able to use low-quality lightmaps since your shadows are in there as well, plus then you get all of the issues of having low resolution lightmaps
Did anyone take a look at how this project works? or has a copy available? the project was removed from Github.
Thanks
I just found out about this and It looks really cool, am interested to hear more.