Flicker

[The Rev]

<a href="http://www.youtube.com/v/VQMTvP7p63w&rel=1&fs=1&hd=1" target="_blank">http://www.youtube.com/v/VQMTvP7p63w&rel=1&fs=1&hd=1</a>

I'm getting a strange flicker between my 2nd and 3rd key frames, as if the light is switching on and off.  I have only one global light, set to (Cos/2+)^2 and 16 Spec Pow.  The only change is the lights X & Y positions between the frames, and an increase in depth fog.  The light sliders go from being at the far left to being at around the 3 and 6 positions (if you assume the sliders go from 1 to 10, left to right).

Any help is appreciate.  Thanks.

The Rev

[Jesse]

Hi The Rev,

i send you a private message, would be great if you could send me some data so i can solve this!!!

Cheers!

[Jesse]

Thank you for the quick response, the reason is that the iteration count is changing very much and because the coloring is spread between min and max iteration count, the colors are cycling very strong.
Plus the inside colors have a strong gradient, so it flickers the faster the faster the iteration count changes.

You could fix the iteration count at the highest value, what slows the first images down... but i fear the total time needed is to hugh anyways 

So it may be faster if you wait until i implement a faster antialiasing method!

And i also can not predict if all the lighting changes can lead to some unwanted behaviour in later frames... would test it in the preview first.

The animation itself looks very promising, no doubt about that 

[The Rev]

Thanks for checking into this for me.  I'm confused by one thing, tho.  The flicker problem begins about half way between keyframe 2 and keyframe 3, and the max iterations for both keyframes is the same, 60.  It then continues to occur between keyframe 3 and keyframe 4, where the max iterations DO change from 60 to 600.  Shouldn't the problem occur only between keyframes with different iteration max/mins?

Thanks,

The Rev

[Jesse]

This is because the bezier interpolation calculates values out of 4 keyframes to get smooth value gradients,
so the interpolated value changes earlier.

I just wonder why the inside colors are cycling, because they should only be spread between zero and bailout?

But the whole coloring, especially in animations, has to be set to a more solid ground... just wondering how.
A problem is that i have only 15 bits i have to press the informations about the color in, with smoothits that
can vary so much and the different formulas this is not easy.

[The Rev]

Do you think that using the Linear rather than Bezier interpolation would reduce or eliminate the problem?  Or just confine it to two key frames?

The Rev

[hobold]

Linear interpolation would confine it, but not fix it.

[Jesse]

Like hobold said, linear interpolation just would reduce the changes between the keyframes where the iteration count varies.
And it makes not really nice pathways or turns.

To the color changes:  the inside colors are the Rout values normalized between zero and bailout, and when the iteration
count changes then the Routs are changing stepwise... so that is the exact explanation.

[The Rev]

I see.  Well, I'm running a preview to see how it works out down the line (I have several iteration changes the deeper it goes).  If I like it for psychedelic weirdness value, maybe I'll finish rendering it.

Do you think that your next upgrade will resolve the issue?  Or should I just try to keep a single iteration setting throughout my animations from now on?

BTW, thanks for all you help!

The Rev

[Jesse]

Keeping the iteration count fixed would be the easiest way, or if you dont need a color gradient for the inside,
just make only one color so cycling would not change it.  The outside coloring still do, or you have to set the
object color sliders in the lighting window according to the shown history between the sliders.
In case of such big changes some changes would remain i guess.

I would not bet that i can solve it generally, not without increasing the memory consumption and storing the color
information as floating point value.

Anyways, i think that animation is to awesome to make a psychedelic version out of it... but i would stay without
antialiasing, maybe with smoothing normals, so the rendering time is in the range of finishing it without regrets.

Edit:  hmm, maybe you would regret it when you did no antialiasing and the result is not worth it... but if you
        choose the antialiasing option, then an update with a faster function is necessary.

[The Rev]

It's all part of the learning process, I guess.  Fortunately for me, I'm obsessed, so I'll learn faster.

Thanks for all your help, and for all the work you've put into this amazing program.  Let me know if you ever make a version for sale.  I want to be the first customer.

The Rev

[Jesse]

Thanks fot the kind words, i don't plan very much in the future, just from step to step.

It takes much time for me to develop any further, so i would not expect so many changes that i
could make an enhanced version for cash or so.  Not to mention that this is not my plan, i even
not expected that the program would get much attention...

[Madman]

I think there's double money here Jesse, first in selling your program and secondly in organizing rehab sessions for Mandelbulb3D-addicts! (Like me..., when's the first session???)

[The Rev]

Totally true! LOL

The Rev

[lenord]

It is rare indeed in our time to see something of quality like MB3D freely given away, for that I am deeply grateful Jesse. I'd be right behind The Rev to buy a copy if it weren't free and I wont Buy much of anything after the crap with Pandromeda/MojoWorld, I'd be right behind Madman too, there is a dire need for a 12 step program for MB3d addicts and it's growing every day.