Random "work in progress" shots

category: general [glöplog]
Abductee: very nice materials. Edges are beveled? (left object, where the stand and object meets)
added on the 2015-11-11 13:32:13 by leGend leGend
pandur: what is te software?
added on the 2015-11-11 21:49:08 by elfh elfh
BB Image
added on the 2015-11-12 01:40:10 by starlight starlight
Abductee, preparing for Under Construction party or just some doodle?
added on the 2015-11-13 14:24:56 by EvilOne EvilOne
EvilOne, Zavie: you can see closeups in http://www.pouet.net/prod.php?which=66494 ;)
leGend: no bevel in the geometry. I think it's an artifact from some time ago, from before we could have sharp normals and no aliasing at the same time.
added on the 2015-11-13 17:00:10 by cupe cupe
cupe: yes, I was going to point out, but restrained myself "to be nice";-) Averaged normals should be something of the past in real-time rendering. Definitely not realistic. Still it is so easy to fall into them.
added on the 2015-11-14 03:05:18 by tomkh tomkh
@starlight: that's a very slow putpixel. You should lock/unlock once per frame just before SDL_Flip (although I never understood if I need these lock/unlock functions in the first place). Also, better to fill an array with pixel data and then give it once to SDL instead of calling many times putpixel.
added on the 2015-11-15 18:56:16 by Optimus Optimus
software render effect. use FPU for this. runs decently speed. 1024x512 resolution.
thought i would share, since im not sure if this effect will ever be used.

BB Image
added on the 2015-11-17 16:46:23 by rudi rudi
BB Image
BB Image
added on the 2015-11-19 14:06:42 by 1in10 1in10
1in10: is that a christmas decoration or something? :)
here's ours: http://www.vogue.co.uk/news/2015/10/26/claridges-christmas-tree-burberry-christopher-bailey
added on the 2015-11-19 14:34:34 by smash smash
its a diy crystal lamp.
2nd try.
first one was done on early led lamp which broke after under 100 hours of use.
next will be first spray painted to give it more depth.
added on the 2015-11-19 14:40:34 by 1in10 1in10
A couple of early drafts on the Offscreen Colonies soundtrack: (these were done with non-64k tools)
https://dl.dropboxusercontent.com/u/3417034/music_pre/64k/off/64ksketch.ogg (still looking for the chord progression here)
https://dl.dropboxusercontent.com/u/3417034/music_pre/64k/off/64ksketch-001.mp3 (got the bass progression, added some drums to show the tempo, still looking for the chords)
https://dl.dropboxusercontent.com/u/3417034/music_pre/64k/off/64ksketch-003.mp3 (changed the instrument, found the melody I was looking for)
https://dl.dropboxusercontent.com/u/3417034/music_pre/64k/off/64ksketch-004.mp3 (started experimenting with the second half of the tune, had a lead that got cut later)

And then I moved it to 64k:

Final tune setup:
BB Image
added on the 2015-11-19 23:58:54 by Gargaj Gargaj
Oh dear I forgot to post this when I was posting about the X Marks The Spot drafts.

Proudly presenting "X Marks The Spot (Low Floating Point Precision Remix)": https://dl.dropboxusercontent.com/u/3417034/music_pre/64k/asmintro/music_output_data.ogg :D
added on the 2015-11-20 00:11:41 by Gargaj Gargaj
That's interesting Gargaj. Is your 64k synth based on buzz machines or is that just what you use to compose your songs before you move them to 64k?
added on the 2015-11-20 00:18:31 by drift drift
This "Low Floating Point Precision Remix" sounds heavily "detuned", it's on purpose I assume? I have mixed feelings about it to be honest. Nice to see "Offscreen Colonies" tune development though.
added on the 2015-11-20 00:21:28 by tomkh tomkh
drift: it's Buzz-based, yeah. allows a lot more creativity imho :) the non-buzz drafts of the OC track were done in Reaper with Massive, and I did the snare layering in there too.

tomkh: nah, it was just a bug that the synth didn't force high enough precision so if the intro itself ran in lower, the synth got horribly detuned :D
added on the 2015-11-20 00:29:24 by Gargaj Gargaj
Hmm, yeah, funny effect. Maybe that's exactly what happened to latest KK intro in high pitch tones ;)
added on the 2015-11-20 00:34:04 by tomkh tomkh
Thanks for the info Gargaj that is interesting and quite a clever idea. It totally makes sense to have that level of modularity and flexibilty. Kinda future proofs the synth with replaceable modules. I keep having thoughts of making a size synth but then I just worry how much time it would take when I already struggle to make 1 production in a year using existing tools.
added on the 2015-11-20 00:49:07 by drift drift
Anyway what this thread is about...
This is the pentagonal bokeh that I decided not to use in my 32k prod because it ran too slowly.

BB Image
added on the 2015-11-20 01:20:16 by drift drift
that's a beauty
added on the 2015-11-20 02:41:21 by 1in10 1in10
indeed - looks really nice
added on the 2015-11-20 13:37:25 by Canopy Canopy
The bokeh is crazy-good, what kind of technique did you use?

Also inspired by the intro in question (xx), I finally did some SSS experiments this evening:
BB Image
BB Image
added on the 2015-11-21 00:30:53 by noby noby
Noby: The DoF is pretty naive, it was my first time trying coding DoF and I just did it in a way that is simple but very slow. The CoC is calculated from the depth buffer like most DoF I guess.

The actual blurring is what I believe is called gather as scatter. It does a circular sample using 2 nested loops, the inner loop samples in a ring around the pixel being sampled, there is a variable for samples per ring (higher = better = slower) then the outer loop scales the rings from the pixel to the blur size, again more rings = better and slower. There is a variable that biases the rings towards the outer circle size so there are more samples further from the pixel center which kinda looks nicer. Oh there is also a small amount of noise added just to jitter the samples a bit. This is what the demo uses as it is released with 3 samples per ring and 3 rings per pixel. It's all drawn to a half size buffer and the CoC is stored in the alpha channel. It gets mixed into the original image using the alpha/CoC as the mask.

The penta shape just has a hard coded look up table of a pentagonal shape, problem is you need many more samples to make the penta shape look crisp. I think the picture I posted above is like 64 samples per pixel and the released intro is only 9 samples per pixel for the circular shape.

I need to do some more research to see what the modern DoF techniques are. I think some people are just splatting out scaled sprites for each pixel. I guess the fill rate on modern gfx cards is good enough to do that now?

The SSS on the water is hilarously badly faked but its very cheap/fast. Yours looks cool, are you actually marching through the objects to calculate their thickness?
added on the 2015-11-21 05:26:51 by drift drift
BB Image
BB Image
added on the 2015-11-21 05:57:51 by MuffinHop MuffinHop