Welcome to Fractal Forums

Hi,

this is my first video test about, fragmentarium camera interpolation from 3dickulus.
As he requested an animation hahaha, i've done one. (fastly during last night)

But please don't be harsh with me, i know this is a draft result with a lot of mistakes and also very short ( 20 seconds )
i didn't really had time and ressources to do more.
I also hesitated to share it, but as it was ok on vimeo!..so let's get go ...

Here the link, for who have 20 seconds hahahaah  :embarrass::

Don't FORGET TO ACTIVATE HD ... otherwize it's more than ugly !!

http://vimeo.com/69771975

some details :

used:
- used 8 keyframes ( interpolation was done pretty smoothly )
- gi renderer with 500 subframes per frame ( approximatly 35 secs rendertime perframe )
- render size 1200 x 600
- used a virtualdub to x264  encode to add some credits and sounds.


Also now , inside fragmentarium , we have a good tool to create some animations.
 :beer:

 O0 :beer:

HAhahahaha !:D  :beer: :beer: :beer: :beer: :beer:

hmmm looks a little bumpy  :-\ is that just internet playback? or is it the actual rendering?

what do you mean by bumpy? if you are talkin about the camera path, it's perfectly smooth ... don't worry about that :)
if there is some jump in the video it's probably because of the encoding ... or so ...

yeah I hope so, looks like a double frame bump, if it looks proper when you play it on your machine then great, if it is a double frame bump then I have to dig into spline routine that generates the points for each frame, they are precalculated and stored in an array so I should be able to dump the list of points and see if there are duplicates or just reeeeally close together.

yes you're right , i've difficulties to play it smoothly on my machine, i thought it was video compression ... :sad1: anyway i've sent u a link to smaller video to check on your side.
but i suspect also, my multitasking to have some impact on this. need to do a real test in controled conditions.

again, hard to tell, not knowing where your keyframes are, the two animations I did were 900 frames and didn't appear to bump, maybe it's the resolution of the spline routine using float, single precision, my video card is a little old and doesn't support double precision so maybe using double inside the spline routine will give better accuracy even if it gets converted to float anyways  :angry:

I'm pretty sure it's not a big deal  :D
i've just sent to yu the lastest compiled  windows binary and the frag i used for the animation if it can be helpfull  ;D

thanks for the frag and exe

I think I've found it, rendering two frames identical, single step through the frames and you should see two the same

I load the mandelbulb-anim.frag and render from those settings, I see frame 54 and 55 as the same 60 and 61 are the same 118 and 119 are the same  :sad1: but it's not where the controlpoints would be  :fiery: they are 1 62 125 188

 it's somewhere in between but I'm not sure until I dig into this a bit further or until someone with better coding skill points out the cause or my mistake  :-\

solution for right now is to hunt down the duplicates and delete them before making the animation from the frames,
oh pardon me, this is not a solution! it's a mediocre fudge! not my favorite flavor of fudge!

hmmm this may take a while  :sad1:

Hahaha ! it's never easy as we want ;D , don't blame you i'm pretty sure yu have the skill to find the solution, at this time a mediocre fudge is ok for me, better than nothing  :D
Don't focus too much take some distance it can help hahahah ! O0
anyway, if you need some help, i'm sure you'll find some ... sorry that i can't do more on my side :tease:

thanks, it'll come to me, I know  :tongue1:

you can do one thing...

load the mandelbulb-anim frag
open the animation window
in the "Frame" box type in 53
then type in 54

when I do this the number stays at 53 so maybe it's not my hack but has revealed an underlying bug in the interface

record the animation frames (max sub frames at 1 for speed) and check 52 53 54 55 or try to find the dups

let me know the result

 :D I think I was right, single precision error, changed things to be double everywhere I could find that had anything to do with calculating frame

I won't believe it's fixed until someone else gets the same results as mine... no more duplicate frames!

and key frames are aligned better in the controller view.

HAHAHAHAHAH Excellent 3dickulus  :beer: :beer: :beer: :beer:
Well done  :D

i'll give a try asap.

here the link to the binary with your lastest code.

http://ge.tt/57lYz5l/v/0?c

thanks, I'll reference it in the thread.

Hi 3Dickulus.

I've taken some time to build another test animation, i had some hard time to encode it smoothly ! but everything seems running perfectly here with your lastest code.
here the link if you want to check.

This one is based on the regular de-raytracer, i've used the same keyframe as previously with the bulb.
i need to re-check on my side with eiffie gi render ...

http://ge.tt/80Yqi7l?c

wow that's quite the bulb! anim looks nice and smooth great!

I'm curious about how the interface feels compared to the original?

Hahaha! this is just a draft test nothing nice in that :p except the smooth camera move  :beer:

To Answer to your question here a screenshot of the win gui, As you can see the sliders for me are an issue. they are very small and difficult to select.

EDIT: I think these changes come from the move to Qt 4.8.5, because i've the same result by building with the current fragmentarium sources. is there a way to change that?
-anyway, it's not a big deal right now ;)

one question, since i can not test it out, if i understand correctly, right now just camera pathes are animatable?!
what about ANY parameter in a keyframe to be interpolated? :(

Yes you're right ! for now only camera path !

i've asked also for dof and focus, the plan there :
http://www.fractalforums.com/fragmentarium/cat-mull-rom-spline-paths-cameraparameter-keyframes/msg63315/#msg63315

HAhahahahahah ! i've solved my issue ... dunno why all windows styles have the same problem.
i've found the way to force a style  :beer: :beer: :beer: :beer:

by adding these lines in main.cpp


 :D :D

hahha now i've usable workspace...

hmm, nice, as far as i see there are now 3 dudes developing the program, be sure to work in the repository ;)))
to make it later available

that looks much better! Qt is nice that way, the style code is very hackable as you may have noticed from the items on my web site.

hahah yes but i'm not a programmer  ;D

yes 3dickulus i've seen that ;) nice work by the way

my github page says only 2 contributors (me and he), Syntopias github page says 1 ... who's the third? have they forked/branched? got a link?

I will only put code into the repository when I have a working prototype, sometimes that will be a brute force implementation of an idea to be cleaned up later but it has to at least function and be useable, the fork I started has a couple of small bugs but it works and hasn't crashed my machine or segfaulted on me and with input from Vinz I'm pretty sure it won't crash a windows box, btw, tnx Vinz :beer:

I have visible paths but they're kinda useless until I can get depth cues working and then, well, you are going to want to point and click to add control points and twiddle them around with the mouse to edit them and... well I'm not sure I'll get that far... in time, we'll see

can anyone demonstrate how to get GLSL raytraced object surface coords into the glDepthBuffer ? I'm on the right track but I think it's a bit more complicated than mapping the "dist" variable in the DErt to gl_FragDepth

I would just put the depth in the buffer and recalculate the coords from the camera position and ray direction. (Not really sure how you are using the info?)

I would like to have the GLSL object occlude the spline paths so...

render object, store depth buffer values, render splines... sounds simple enough

the depth buffer has a range of 0.0 - 1.0 while near/far is 0.001 - 100.0 (probably excessive but can be adjusted)
linear Z buffer = ( Z - nearZ ) / ( farZ - nearZ ) and non-linear Z buffer = ( 1/Z - 1/nearZ ) / ( 1/farZ - 1/nearZ )
OGL is non-linear I think ie:more resolution when near less when far

need to adjust the hit.xyz? values in a hybrid sort of way with xy as screen coords and z as depth buffer value?
or the distance/length requires transform and scale from eye to screen matrix ?

I'm not really sure about the GLSL side of things, input would be appreciated.

just played around with that, great idea!, should be an option in preferences
you don't really need the include code (on linux) just one line is enough to do it