Page 1 of 1
Posted: Thu Feb 15, 2007 8:59 am
First off this program is amazing. I use nothing but mac, (since I work for Apple
but I was having to dig out my old PC just to run Apophysis to do fractals. I'm so glad that this program has been created, it is far more superior to Aopohysis in every aspect... except for one...
I always render flames in PNG with a transparent background (thanks for the bug fix btw). When I rendered in Apophysis at the highest settings, the flames came out very clean. When I render in Quadrium, there is "dust" and the images don't appear to be as smooth.
I'm rendering at 600dpi, alpha channel, antialias 4:1, 1920 x 1200.
I have played with the gamma and rendered the same image out at all different settings and jpg seems to be the best (which is confusing because shouldn't PNG be better?). I know my settings are wrong somewhere because I've seen examples from Quad and there all perfect.
Also, I do take them into Photoshop for clean up, but that doesn't seem to matter much.
Posted: Thu Feb 15, 2007 9:44 am
There's a couple of things to try.
The single best thing is to turn up the "Subsampling" slider - it will slow things down, but result in an improved image (it's basically like turning up antialiasing without the horrendous memory requirements). Note that you can crank it up to values higher than the default maximum of 3 by typing in a value.
The next thing to do - and this is trickier to work with - is to turn up the "Blur" slider. The problem with "Blur" is that it is a pixel level manipulation, which means that the effect of blurring 1 pixel at a 72 dpi is going to look different than 1 pixel at 600 dpi (with high antialiasing). It's worth at least giving a try to see if it help.
You can also try the ZRGB rendering mode instead of the RGB - it will give slightly different results, but it handles a few curves differently.
I've been kicking around a few "adaptive resampling" techniques, but those aren't ready for prime time yet... At one point, I had the thing plot little Gaussian spheres instead of single pixels - the result was just a blurry result with not much better "fill in" in the faint areas, and loss of detail in the sharp areas.
Posted: Tue Feb 27, 2007 4:00 pm
I'll give this a look later on but I have asked a similar question in another thread.
I spent several hours playing with a flame yesterday and no matter what the image still comes out very noisy - Banging Subsampler to the max cuts out a lot but not enough
I have looked at all Flame renders on the Deviant Site and have not seen any clean renders in this respect. My very first Flame with this program was fine but itreally depends on whatkind of flame I guess.
Apo flames a lot cleaner
Am having a real hard time trying to firure this program out and reitereate what I said on Deviant about it not beong intuitive - you may yet prove otherwise.
When I did geometry in school you would plot points on the x and y axes and connect them to form a triangle. To a triangle you can apply sevral calculations to alter its form via trigonometry.
All I see in Flame are F's and not triangles and something like an alpha-cryptic wheel to alter position and orientation of the F's - more like a PC game I played a while ago than anything you could apply any logical operation to
The sliders have lots of variables for the most part only adding more or less noise to the F's from whay I can see.
These are among but a few of the observations I have had on Flame so far that has left me screaming for in depth tutorials.
When I see something like the julian forms I created in a matter of hours with Apo yesterday (and I have only been at this whole deal for around 5 days now) created with Flame and without all the noise, plus a walk through with screenshots of the setting as to how it was arrived at, perhaps then I will take this program seriously. Right now I am getting far superior results with the Free Apo than what I ended up paying 45 euros for on MAC
Posted: Tue Feb 27, 2007 6:54 pm
I guess it's worth mentioning that on some flames 2:1 gets much smoother results than higher anti-aliasing. Always worth a try.