On 2/5/21 2:28 AM, Thomas de Groot wrote:
> Op 04/02/2021 om 13:56 schreef William F Pokorny:
...
> 
> Ah, yes! Luminous bloom! Something I have long intended to use indeed
> 
>>
>> Attached a result. I expect it can be better tuned, but I'm happy that 
...
> 
> Not (yet) convinced. Like Mr says, looks to much like blur here and 
> everything is blurred.
> 
> I have the intention to do also a stochastic render of the scene. I 
> /think/ that would give a better result. 

Yes, that would be interesting to see. With the stochastic techniques 
I'm always wondering how much is a better render result and how much 
looks better because one has introduced noise. And if that last true, 
even in part, might we add noise by some more efficient means. Anyway... 
Always a thousand ideas.

I've been playing with more ideas using your image. Attaching three 
images. In toBloomOrOther.jpg showing your original to my already posted 
bloom filter image in the top row. In the middle row the bloom filter at 
about 1/3 the aggressiveness of the top row. In the bottom row not 
really bloom, but more adding noise by regional sampling about each 
pixel. Less blur in the bottom two rows, but still maybe too much to 
tastes.

While at that, Mr's question about adding more contrast knocked 
something loose in my head and I had the thought, "what does average do 
with negative weights...?" Well! Interesting stuff - about which I've 
not completely wrapped my head.

You can use negative weights. If you get the balance right you can get 
an image with more contrast with my bloom filter set up. Using:

      #declare PigmentMap00 = pigment_map {
     [-1.0 Pigment1 ]
     [-0.7 Pigment2 ]
     [-0.6 Pigment3 ]
     [+0.5 Pigment4 ]
     [+0.4 Pigment5 ]
     [+0.3 Pigment6 ]
     [+0.2 Pigment7 ]
     [+0.1 Pigment8 ]
}
#declare PigmMerge = pigment {
     average
     pigment_map { PigmentMap00 }
}

I get the Contrast00.jpg image, which isn't traditional contrast, but 
something more along the lines of tone mapping. Without even trying! I 
find it amusing it's possible to stumble my way into such functionality. 
:-) Aside: I shrank the image size because it got large even as a jpeg 
due the detail popping out - the detail jr wanted to see and probably 
still can't. ;-)

If you get the balance for contrast slightly wrong, other interesting 
things happen. See Cartoon00.jpg. The only difference is the -0.6 weight 
above was instead +0.6.

Creating these last two images is fast supposing the eight image 
pigments into the average function already exist. Whether with effort 
and exploration techniques using negative average weights could be made 
more finely controllable - in other words, truly usable - I don't know.

So many things to play with and so little time.

Bill P.

Post a reply to this message

Attachments:
Download 'cartoon00.jpg' (269 KB) Download 'contrast00.jpg' (288 KB) Download 'tobloomorother.jpg' (198 KB)

Preview of image 'cartoon00.jpg'

Preview of image 'contrast00.jpg'

Preview of image 'tobloomorother.jpg'