Search Issue Tracker

Won't Fix

Votes

0

Found in

2019.1.0a13

Issue ID

1114566

Regression

No

LPPVs sampling Light Probes which store intense direct light appear blocky and somewhat corrupted

Progressive Lightmapper

-

Repro:
01) Download and open the attached repro project LPPV_SamplingWeirdness.zip
02) Auto should be enabled. Allow this to complete a bake)

Observe:
The Scene has two examples, one with a high vertex density cuboid (made of multiple sub cubes) receiving a single LPPV which encompasses the entire group. The other example uses a cuboid from a single, scaled cube which also samples an encompassing LPPV.

Direct light from some coloured point lights is sampled into the dense grid of Light Probes which is then passed to the LPPV. The result on the receiving objects is blocky, and does not resemble the uniform patches of light we would ideally like to see from the point light source.

I imagine this might be because the 2 band LPPVs are ringing when sampling the 3 band Light Probes. This is then being sampled into the 3D texture, which introduces further sampling artefacts that appear as inconsistent intensities where darker regions or bands have been used. At least this kinda makes sense in my head.

Expected:

Light interpolation across the LPPV should ideally be smooth, without noticeable blocks. It should also more closely match the the smooth, uniform shape of the falloff from the point light source. What can we do about this? Is using 3 bands for LPPVs a non-starter?

  1. Resolution Note (2020.1.X):

    This limitation is caused by hardware texture interpolation. High light intensities make this issue worse. The upcoming improvements to light probes should make this issue obsolete.

Add comment

Log in to post comment

All about bugs

View bugs we have successfully reproduced, and vote for the bugs you want to see fixed most urgently.