Left: preview - point light moved away from the flame mesh that I thought was causing the problem.

Centre: SSS pass

Right: final render

The point light is using inverse square falloff and its intensity is set to 2%.

The candle is using the milkman Andy material

The problem only appears when the light is in very close proximity to the candle top - which it will be as it represents the candle flame

I'm fairly sure I've seen a similar effect somewhere before. Is there any known way to avoid it, workaround, or whatever ?

(The scene is zipped and uploaded to post 52 of the Hivewire3D Poser Firefly candles topic if you want to reproduce the problem)

I am assuming that the point light is within the flame mesh.  Make the point light either the same size or a bit larger than the flame and see what that renders.

No. That's what I originally thought the problem was, so I moved it out - you can see it top right of the preview, well away from the flame mesh.

It doesn't look quite like polygonal self shadowing, a side effect of Reyes rendering exacerbated by either too low a shadow min bias, or Poser phong smoothing, or AO.

You can try turning off smoothing in Firefly render settings, raising your shadow min bias in properties of your raytrace shadow light, and or applying subdivision to the mesh(es) in lieu of smoothing. Usually a level of subdivision helps in conjunction with smoothing off. Back in the day before SubD in Poser I would fiddle with shadow min bias and arrived at 0.350 as a value that helped. Values below 0.200 are a sure culprit if you are using smoothing (which was really necessary back then to eliminate the faceted look of many models)...

But caveat, like I said it doesn't look quite like the artifact type I described... which could best be described as a sort of mesh XRay appearing over the surface of the afflicted model when rendered. A bit hard to tell in your image.

"Mesh X-ray" is exactly how I'd describe it (the candle originally had a well in the centre, hence the concentric circles).

My min bias was set to default 0.8 for the original screenshot and for this one. Here I have Smooth polygons disabled but get the same effect (the light is closer than in the original image)

Reducing min bias to 0.35 or 0.2 made the mesh line 'shadows' slightly thinner, but ridiculously low values reduced the mesh line thickness a lot, although it had no effect on the the grey eye-shaped area around it.

It appears that I can get rid of the mesh X-ray part of it by reducing light intensity, provided the light isn't really close to the candle. If it is then the intensity I have to reduce it too isn't sufficient to simulate the candlelight, which is its sole purpose.

I also tried using the other falloff modes. I think inverse square is the most prone to the mesh X-ray effect. But it's also the best for candle light.

With default min bias I still get that grey circle (it's not eye shaped in these renders). The point light is to the right of the flame - you can see it in the preview

Definitely nothing to do with the milkman Andy material - the sequence below uses just PoserSurface diffuse colour. Left to right the point light is gradually dropping lower (it's directly above the small shadow spot in the 3rd image, and drops from roughly top of flame height in 1st to roughly top of wick in 6th).

Inverse square dropoff with min bias at default 0.8

Changing min bias with the point light roughly at the position of the 5th image above has negligible effect on the circular grey shadow

Just for curiosities sake try adding a level of subdivision in the properties... also does the shadow disappear when you turn off shadow casting for the candle?

If it does then the solution would be to remodel the candle with the flame part(s) as separate parented props and have shadow casting turned off in those part(s) properties... obviously not ideal but a solution nonetheless.

As a, perhaps, contentious observation I'll admit that this candle artifact dilemma is rather puzzling in my eyes. You've been working on this (non cowardly) candle flame thingy for how long now? 2 years? Of course not continuously... With a ton of feedback from others also. It just seems strange that this is happening only now, yes?

Subdivision won't solve this problem.  I have already tried this method and the result is still the same.

primorge posted at 5:57 AM Thu, 2 June 2022 - #4439472

Just for curiosities sake try adding a level of subdivision in the properties... also does the shadow disappear when you turn off shadow casting for the candle?

If it does then the solution would be to remodel the candle with the flame part(s) as separate parented props and have shadow casting turned off in those part(s) properties... obviously not ideal but a solution nonetheless.

I'm using the figure version of the candle (see the link in the OP) which has separate bones for the candle, flame, light emitter4, smoke, and point light. By default I have only the flame and light emitter bones set to NOT cast shadows. If I also set the candle bone to not cast shadows...

So it appears to be the shadow of the candle !

Unfortunately your solution won't work because that's what I already did for the figure. (Just as a double check I ensured that all bones EXCEPT the candle bone were NOT casting shadows - the problem still occurs.)

primorge posted at 4:40 PM Thu, 2 June 2022 - #4439491

As a, perhaps, contentious observation I'll admit that this candle artifact dilemma is rather puzzling in my eyes. You've been working on this (non cowardly) candle flame thingy for how long now? 2 years? Of course not continuously... With a ton of feedback from others also. It just seems strange that this is happening only now, yes?

Just over two years later I was doing a render with a candle and I needed a flame - that reinvigorated the topic

Don't forget I'm mainly trying to create a candle flame shader - I wasn't concerned with how it was going to illuminate the scene until I got the shader right, which was 10 days ago when I released it as a freebie. I'm using a point light because that's easiest for me - I'm not familiar with the ins and outs of light emitters. Finally, this problem simply didn't show up in any of the test renders or animations I did. The first time I saw it was with this figure version of the candle.

I played with it a bit and found that it disappears if you lower you shadow blur radius on the light.    Taking it to 2 seemed to work well, of course you have sharper shadows,  but still didn't look bad to me when I added a cone to the scene, turned the ground on and tested it.

Excellent !

Reducing the shadow blur radius does indeed seem to get rid of it.

For my next trick I took the CandleV3figure, scaled it down to 10% (a realistic candle size in Poser world), set light intensity to 0.2%, and put it in front of La Femme. Shadow blur radius back at the original 10 here:

And simply reducing shadow blur to 2:

So the high shadow blur setting explains some odd stuff I'd seen before but hadn't mentioned.

I noticed that I have a grey circle on top of the candle in both renders (the point light is at the tip of the flame). If I move the point light forward the circle moves forward too, so it seems to be the same artefact I was getting alongside the mesh X-ray. Moving the point light slightly forward supports this as the circle moves:

It disappears if I set the candle to NOT cast shadows.

So reducing shadow blur radius can get rid of the mesh X-ray, but not necessarily the grey circle. But you get less blurred shadows, and high shadow blur for me fits the candleflame.

Setting the candle to NOT cast shadows gets rid of both mesh X-ray and grey circle, but you don't ANY get candle shadows.

I wonder - if I change the light to inverse linear...

Shadow blur radius would have never occured to me from afar, the advantages of testing the actual object and some lateral thinking... Well done Bwldrd.

I don't think I've ever seen such a problem with cast shadows before, the only reason I know about the min bias and subdivision solution is via a long ago lecturing from Bagginsbill... But that wasn't a cast shadow thing but rather a direct object surface problem.

Being that you've already released the candle prop as a freebie and the overall solution is turning off shadow casting for the candle, maybe update your freebie with a different version? It's obviously the flame part of the prop that's the main culprit... So an update with the flame part separated as a parented part with it's shadow casting off, leaving the candle body as the shadow casting source, would seem to be best. That's what I would do at least.

... Of course that would bork the morph targets. Probably pretty easy to extract the default flame part and establish a vertex order and iterative generate like vertex order extractions for the morph targets. Then again they're not very complex morphs, relatively. You could just recreate the morphs by hand.

Oh. I missed one of your posts. Ah well. Best of luck with that, my time is extremely finite for my own projects so I'll have to leave it at that. Sorry I couldn't help.

Thanks for the replies primorge. I don't have any problems reworking things, or even going back to square one and restarting from scratch. I do that a lot. :oD