higher resolution

Features you'd like to see in future versions, gripes, ideas..
Post Reply
rygoody
Posts: 3
Joined: Mon Sep 15, 2008 9:33 pm

higher resolution

Post by rygoody » Mon Sep 15, 2008 9:55 pm

I have a model that to get the resolution in shadows that it needs I need to set the resolution to 1". But this makes the problem that, sketchup then complains about requiring a texture >1024 when computing the lightmaps. Then all the large surfaces in the model aren't lit, because I assume those surfaces needed a texture > 1024.

I imagine I could go through and chop up the large surfaces so that it can compute it at this resolution. But really it should be able to automatically handle this.

Also, this really needs to be multithreaded, so much CPU power going to waste.

User avatar
AdamB
Site Admin
Posts: 2294
Joined: Wed Apr 16, 2008 10:11 am
Location: Brighton, UK
Contact:

Re: higher resolution

Post by AdamB » Tue Sep 16, 2008 11:46 am

Yes, both good suggestions and ones we'll get to in time.

I do question why you feel you need 1" resolution (assuming this is a building). Also, if you have large areas that perhaps don't need such fidelity you might use the keyword "_lightup(lowdensity)" on those materials to reduce lightmap usage.

As to multicore support, barring any unforeseen showstoppers its in the next version (v1.3).

Adam
Developer of LightUp http://www.light-up.co.uk

rygoody
Posts: 3
Joined: Mon Sep 15, 2008 9:33 pm

Re: higher resolution

Post by rygoody » Wed Sep 17, 2008 12:15 am

The building has a curved array of small windows on it, about 12" x 12" each going across an arc.
With the low resolution sampling each glass pane did not get enough samples on it to accurately shade itself. It looked like each window was randomly shaded. 6" sample size worked fairly well. 3" is where it started to become how it should, but 3" had the same error as the 1". So either way it woulda still needed to support larger textures.

Although, is there a _lightup(highdensity)? I guess that would of fixed this.

You know what would be really sweet is if the algorithm could automatically vary the size the of the samples. Or what if it didn't even store the lighting data in an image file? But rather just in an array that could be read back and applied dynamically, so then a single surface could have variable sized samples on it.

User avatar
AdamB
Site Admin
Posts: 2294
Joined: Wed Apr 16, 2008 10:11 am
Location: Brighton, UK
Contact:

Re: higher resolution

Post by AdamB » Wed Sep 17, 2008 10:26 am

rygoody wrote:Although, is there a _lightup(highdensity)? I guess that would of fixed this.
Just committed a revision (r679) to support an additional material name keyword _lightup(highdensity) that doubles the normal lightmap resolution for faces using this material.
rygoody wrote:You know what would be really sweet is if the algorithm could automatically vary the size the of the samples. Or what if it didn't even store the lighting data in an image file? But rather just in an array that could be read back and applied dynamically, so then a single surface could have variable sized samples on it.
Yes, we could store a 2D array of RGB triples and then for each pixel work out which samples it subtended in the array and if it was <= 4 samples, interpolate between them or if too many samples switch to a lower resolution 2D array of RGB triples and so on... :-)

Adam
Developer of LightUp http://www.light-up.co.uk

Post Reply