A really great way for adding textures to your paints and/or demos...
category: general [glöplog]
I'm not sure if you already known about this method or not, but I've not seen any demo with this effect yet, so, I explain.
It is a well known method in photography for "selective unsharp mask" alike effect. The steps in photoshop are:
1) Duplicate the layer of your image
2) Apply in the upper layer a filter-other-high pass to a little value (for example, radius 0.8)
3) Change from "Normal" to "Overlay" mode the upper layer.
Now you have a very similar to unsharp mask effect, and you can see and modify where it is applied.
But, of course, you can change the step 1) by "1) Add a new layer with any texture", and the results are impressive, as examples:
or this updated version of the giger pig:
It is very useful for photoshop, but it can be easily coded also. I've already coded it for my software rendering demosystem and it can be easily done by hardware. The high-passed textured could be stored directly in a .jpg and the Overlay filter in photoshop works this way for every color channel (supposing colors are floating point 0 to 1 values):
if (src_color<0.5) return (src_color*dst_color) else return (1-((1-dst_color)*(1-src_color)))
It is a well known method in photography for "selective unsharp mask" alike effect. The steps in photoshop are:
1) Duplicate the layer of your image
2) Apply in the upper layer a filter-other-high pass to a little value (for example, radius 0.8)
3) Change from "Normal" to "Overlay" mode the upper layer.
Now you have a very similar to unsharp mask effect, and you can see and modify where it is applied.
But, of course, you can change the step 1) by "1) Add a new layer with any texture", and the results are impressive, as examples:
or this updated version of the giger pig:
It is very useful for photoshop, but it can be easily coded also. I've already coded it for my software rendering demosystem and it can be easily done by hardware. The high-passed textured could be stored directly in a .jpg and the Overlay filter in photoshop works this way for every color channel (supposing colors are floating point 0 to 1 values):
if (src_color<0.5) return (src_color*dst_color) else return (1-((1-dst_color)*(1-src_color)))
Nice. There arent ANY prods that use this? I would like to see it implemented.
That HR Piger image is absolutely fucking nutso great!
Actually a lot of demos/intros use it - it's called bump-mapping :)
The highpassing of the image causes the picture to go all gray with only the delta values of the differences left (see also - Emboss). If you use that to shade your image with some kindof bumpmapesque shading, you will get a much edgier, grainier looking result.
And the above process was also described by Ile in his texgen tutorial.
The highpassing of the image causes the picture to go all gray with only the delta values of the differences left (see also - Emboss). If you use that to shade your image with some kindof bumpmapesque shading, you will get a much edgier, grainier looking result.
And the above process was also described by Ile in his texgen tutorial.
mb someone hasn't read it yet. you can find it in Hugi SE on coding
Gargaj, of course this is similar to bump mapping, but the important is the way it could be used by graphists, directly from Photoshop to the demo. I'm not sure about how the highpass filter of photoshop works, but it doesn't look the same as emboss, highpass looks as using pondered values by distance to radius, maybe similar to gaussian blur. Also, the most important for me, is that bump-mapping (at least the oldschool one I know) is being done by multiply (with or without saturation), and this one is done by Overlay. Visual results are very different, this function is cutted in two parts, only one part is equal to multiply.
*I think I had a little error by not adding the multiply by 2, so, now corrected:
if (src_color<0.5) return (src_color*dst_color)*2 else return (1-((1-dst_color)*(1-src_color)))*2
*I think I had a little error by not adding the multiply by 2, so, now corrected:
if (src_color<0.5) return (src_color*dst_color)*2 else return (1-((1-dst_color)*(1-src_color)))*2
woot. nice. that image looks great, btw.
i wonder why is nanook a fakeuser, it makes me upset as i have fakeusers' posts disabled but nanook does not post such utter bullshit ( lately atleast ) as other fakies do
" it could be used by graphists"
sorry but no one with a clue would use it.
there are even better ways to work out blurred details if needed
but without changing contrast and colours.
actually i dont see any sense in it.
perhaps this answers your question why no one uses it...
coz no one needs to use it.
sorry but no one with a clue would use it.
there are even better ways to work out blurred details if needed
but without changing contrast and colours.
actually i dont see any sense in it.
perhaps this answers your question why no one uses it...
coz no one needs to use it.
fadeout, he explained it in the pig thread.
I never disable fakeusers. This way, I would miss the most fun!!! :]
fadeout!: my fakeuser status is well applied, this was not my main account (even if now I only use this one), and I used this account to fakevoting on some minimalanimal jokeprods. I don't care so much about it anyway...
But doing these steps inhardware would cause the texture to being different on different systems!!
Everone does NOT HAVE THE SAM VIDEOCARD!
Everone does NOT HAVE THE SAM VIDEOCARD!
Everone does NOT HAVE THE JACK TASTECARD!
heulanith: thankyou for displaying utter incompetence.
nanook, so who are you anyway, then?
gargaj: after your claim that this technique was bump mapping, i wouldn't be the one to speak if i were you ;)
kusma: it might not EXACTLY be bump mapping, but the "shading-with-deltamap" technique is pretty close, the rest is just semantics..
reducing one technique to another generic one is not a way to argue, gargaj :) otherwise, we can always dismiss any kind of effect as being boring because it is just pushing pixels.
I'm not reducing anything, i just stated that the techniqe by it's fundaments is rather similar to what people usually mean under bumpmapping.
all these threads hurt my brain
OFF WITH GARGAJ'S HEAD!!!!
seems like a boring effect, at least in static mode. think it will be superattractive and all obvious to the viewer of a demo once that one's getting in action?
:D..
