Welcome to Fractal Forums

I must make a big print with one of my fractals and I decided to render again the flame.
This time I render it with a bigger size: 4724x4724. The first file was rendered 3200x2676.

As you can see in the file attached, the results are different, the details in the second file are different.

Is it normal? I don't remember how and if I post-processed the first image 3 years ago, but I don't think those details are due to some PS work.

Thanks for any tips! 

ifs renderings need much longer renderings when size is increased ... :(

Hi!

I'm not sure I get it.
You say it's the rendering time that can change the final result?

Try to do an equalization in Photo-whatever, it almost always improves. ;) As well as auto-contrast ... I try to apply everything with a blend, it improves a lot most of the times

I tried lots of things in Photoshop, but I don't think I made the first one looking like that in PS.

O probably rendered the first one as jpeg and now I chose PNG. Do you think it can change the final result?

Improves does not mean that you can add missing details; only make more visible what the image already has :(

Jpeg and Png on very large images should not be a big difference.

That's it :)

(my fave filters are KPT equalizer, sharpening after a very subtle blur to remove nasty pixelation)

Ok. I guess I have to dig more in PS.

Thanks for your help!

It's not the render *time* that's the problem directly it's the number of *iterations* performed - this needs to increase proportionately based on the fractal dimension of the flame.
e.g. if you render a fractal at 600*600 that has a fractal (hausdorff) dimension of 2 and render it again at 1200*1200 then you need to increase the number of iterations performed by 4 - whereas with a fractal of fractal dimension 1 you'd only need to double the number of iterations - and for a fractal of fractal dimension >2 then you'd have to increase the number of iteration by more than 4*.
Obviously this does feedback to the time taken since a fractal of dimension 2 rendered at twice the original size will thus require approx. 4* the render time of the smaller version.

Apologies from me though as although I worked out the exact relationship in my RIFS beta formula for UF I've forgotten the details of the relationship between the fractal dimension, increase in resolution (and/or magnification) and required iterations.

David, thanks a lot. Woow! It's all about math after all.

I'm still learning here so I'm not sure I get it. What do you mean be "iterations"? Do you mean the numbers for Filter Radius and Oversample?

Thanks!

PS: I decided to make more test and render my files as jpeg as well with different settings ans sizes.

Most flame software works by essentially repeating a given loop many times i.e iterating the code in the loop - a single pass being one iteration - how to increase the number of iterations used varies from one piece of fractal software to another and in fact some (like my beta formula for UF) auto-magically adjust the number of iterations for you when you change resolution and/or magnification based on the fractal dimension of the object - it's possible that in some software you really do need to just increase a time factor parameter ?
I don't know as I don't use any flame software.

I remembered the formula (I think).

Take an initial render that looks "correct" that is w*h in size (pixels) and at magnification m and uses i iterations (or time t), then:

If we wish to render the same image at twice the size i.e. 2w*2h (pixels) then we need to do sqrt(2*w*2*h/(w*h))^D where D is the fractal dimension.

i.e, we need to change the iteration count to i*sqrt(4)^D or increase the time to t^sqrt(4)^D i.e. i*2^D and t*2^D.

The full formula, allowing for changes in height, width and magnification is: