Topic awaiting preservation: Diffuse map for Earth

Hot damn! I love it! It's big and colourful. Not only that, but the ratio is proper for proper spherical UWV mapping (2048 x 1024). Beats the everlovingshit out of all other Earth maps I've seen.
http://antwrp.gsfc.nasa.gov/apod/ap020305.html


edit: Wanna make a bump map to go with that? http://antwrp.gsfc.nasa.gov/apod/ap010925.html

edit2: How about self-illumination to show population? http://antwrp.gsfc.nasa.gov/apod/ap001127.html




[This message has been edited by warjournal (edited 07-08-2002).]

[This message has been edited by warjournal (edited 07-08-2002).]

Holy living crap...I just came up with the best goddamn render of earth I've ever seen!

Koan 63, written on the wall of cell number 250:
Those who Believe
Can
Those who Try
Do
Those who Love
Live

The poster has demanded we remove all his contributions, less he takes legal action.
We have done so.
Now Tyberius Prime expects him to start complaining that we removed his 'free speech' since this message will replace all of his posts, past and future.
Don't follow his example - seek real life help first.

wow... this is awsome...
I think I'll try to make 'the-coolest-earth-render-I've-ever-made'..

ok, I'm going to need help with the following elements (in 3dSMax, btw):

I'm trying to make a nifty atmospheric sphere around the planet, but opacity changes look not-so-good... anybody know what I can do to make it cool?

what's the best codec/renderer to render something like this so it looks crisp and sharp?

when I get back I'm going to try desaturating the bump map in PS to see if it looks better...

btw, when I come back, I'll post links I found to the moon and the far side....


Code - CGI - links - DHTML - Javascript - Perl - programming - Magic - http://www.twistedport.com
ICQ: 67751342

InI, I mostly do modeling, either for my own edification or for others, so I don't really post my work. Here's one I was working on not too long ago. Same thing from the outside in a progression-type presentation. Another.

I haven't tried the bump map yet, but I suspect the order of the colors will cause some problems. I'll see if I can make a more correct de-saturated version.

I'm pretty sure that I've got the height map fixed. The parts of the world that I am familiar with look good, but I could be wrong. Somebody care to verify?

earthheight_bigfix.gif - 539 k as a GIF. You might want to blur it a touch and save as a JPG. I would be offering it as a JPG, but I wanted to keep the 256 shades of grey as close to the original as possible.

Let me know if it's good or if it's too much out of whack.

Petskull... you will need the maps from this guy http://gw.marketingden.com/planets/planets.html
also take a look at the last image I did in my pong match with ini. If that is what you are looking for I can tell you how to do it.

Another good link is this http://maps.jpl.nasa.gov/



[This message has been edited by reitveld (edited 07-09-2002).]

Nice, Reitveld. How long have you been holding out on us?

Compared to his bump, mine is awlfully close.

edit: Love this about Earth's bump: "Some madmen have also used this data in POV Ray as a displacement map on a very finely divided sphere to produce a "true" 3D version of the Earth."


[This message has been edited by warjournal (edited 07-09-2002).]

Please don't think I'm trying to be an ass or anything, but that's a color map of the earth. not diffuse. ^_^
Diffuse would be a greyscale map depicting how much light is absorbed and scattered for an object.

Here is a link to a lightwave tutorial on making a photo-real earth. some great maps to download too.
http://personal.southern.edu/~dascott/tutorial01/nasa-earth.htm

No...that would be a specularity map. Go have a look at your 3D program...you'll find the area where you can define the actual coolour of your material is called 'Diffuse', making any sort of colouration or picture/image map done on this channel a diffuse map.

Koan 63, written on the wall of cell number 250:
Those who Believe
Can
Those who Try
Do
Those who Love
Live

Yeah, although for some reason I thought specular when I heard the word "diffuse", diffuse light is in fact the light that's scattered in all directions after hitting an object, which is what actually gives an object its color. So it makes sense for a diffuse map to be a pigment map.

I found a great cloud transparency map once for the same purpose, but I forget where it is...

And the left over light which bounces off and manages ot colour objects a certain distance away is called hte radiosity...oh the fun of light.

Koan 63, written on the wall of cell number 250:
Those who Believe
Can
Those who Try
Do
Those who Love
Live

Specularity will show the hotspots of a light by reflecting. Diffuse is how the light gets scattered. They are completely different things. Diffuse maps are generally greyscale like a bump map or a spec map.

3DS Max has it backwards in it's program. ^_^

Edit: here is a nice article posted on cgtalk that explains it better than I can. http://www.cgtalk.com/showthread.php?s=0ccf7f273d56c4ab88a1ac8388ae491c&threadid=11053


[This message has been edited by Rumors (edited 07-09-2002).]

wow, that was one hell of a read.... worth it, though...


Code - CGI - links - DHTML - Javascript - Perl - programming - Magic - http://www.twistedport.com
ICQ: 67751342

As I read that article (and as I previously understood it), you're mistaking the term 'diffuse color map' with 'diffusion map'.

The article linked above describes (correctly) a diffusion map as a grayscale map that controls how much diffuse reflection a surface has at a given point.

However, a 'diffuse color' map is a map of the color generated by diffuse reflection at a given point, as opposed to the color generated by specular reflection at a given point. A specular color map controls how gold the specular highliights are off of a given spot on a gold ring, for example. A diffuse color map would control how green the tarnished bits are (guess it wasn't real gold ...).

Basically, the 'diffuse color' map controls the color at the point. This would be like the color earth map above. The 'diffusion map' controls how much light of that color reflects from that point. One is color, the other amount.

Often, the diffusion map is just faked by using darker or brighter colors on the diffuse color map. That doesn't work well for GI calculations, though (a high diffusion dark color is not the same as a low diffusion light color when you're determining how much light bounces onto nearby surfaces).

While most apps don't have both maps for diffuse reflection, most do support both maps for specular reflection. The 'specular color' map controls the color of the specular highlight at a given point, and the 'specularity' map controls how much light reflects from that point.

There's also the 'gloss map', which controls how 'tight' the specular highlights are. It's sometimes inverted and called a 'roughness' map (since rough surfaces are very non-glossy).

To add to the confusion, different apps often use different terms for all of the above.

Max 4's names for the maps I discribed above are:
Diffuse (diffuse color)
Specular (specular color)
Specular level (specularity)
Glossiness (gloss)
It does not support the diffusion map. If it did, it would probably call it 'Diffuse level', to match the corresponding specular level map.

It seems that with these definitions, a diffusion map is easily simulated by making it a black and white layer in photoshop and multiplying it with the diffuse color map.

Yes, that would be one technique for the 'faking it' I described above, where dark colors are used to simulate bright colors with low diffusion. Works fine for most situations, GI being the main exception.

Another solution is to use a 'dirt' plug-in. Most of these work by darkening the color according to a greyscale map, to simulate the surface being more or less dirty/smudged at different points. This is close enough to a diffusion map to give much the same effect.

Re-reading the linked article, your technique is also what the author is using. She inverts the specularity map as a base, tweaks it, then combines it with the diffuse color map to fake a diffusion map. The diffusion channel she's using exists only in Photoshop, not the 3D app (Max, looks like). So you could totally fake the effect of the diffusion map by just painting your diffuse color map really really well. It does look like using the map would help a lot, though, since you can combine your specular map and the 'scratches and scrapes' part of your bump map and work from there (since both specular level and scratches will alter the diffusion at a given point).

[This message has been edited by Das (edited 07-10-2002).]

Actually lightwave has a diffusion channel, and being the app that the author is using, she is indeed putting a diffuse image in the diffuse channel. She paints the maps in photoshop and saves them out as seperate images.

You can darken the color map but it still won't affect how light scatters off the object, that's why the diffusion channel is important.

Is she? She should really say that somewhere in the article. She mentions both Max and Lightwave, but not what app she's using.

Regardless, having the seperate map is not really that important. Read Mauritus' reply

Since the diffusion map can be baked into the color map, it doesn't matter if you actually have seperate maps in Lightwave or one composite made in Photoshop. With the photoshop technique, you're just 'pre-baking' the diffusion map into the diffuse color map. While having a diffusion map makes things more convenient, I'm not convinced it offers any concrete advantage, outside of GI.

I do think her article is very informative on just what a diffusion map is for, and why they're important. I just don't see the need for using it in the app vs. creating it in Photoshop and combining it into the diffuse color map.

[This message has been edited by Das (edited 07-10-2002).]

Ah, even more relevant, the author agrees in a later repley that combining the diffusion map with the diffuse color map in Photoshop is effectively identical to using both maps in Lightwave. She prefers using the two maps in Lightwave.

It does make things more convenient having the diffusion map in Lightwave, since you don't have to switch to Photoshop to change it's intensity, etc., then save it, then update it back into Lightwave. You would just adjust it's intensity in the surface editor.

But it doesn't really make a difference in the final image which method you used.

[This message has been edited by Das (edited 07-10-2002).]

So, what's the point of a diffusion map then? To simulate radiosity manually?

POV-Ray should support this =)

It is possible to combine it in photoshop, but I still don't think it looks right. I made some (admittedly poor) samples to show what the diffuse channel does.

plain color map http://www.badsun.com/images/DV_Samples/Diffusion/1colormap1.jpg

render with plain color, no diffuse http://www.badsun.com/images/DV_Samples/Diffusion/2nodiff.jpg

render with plain diffuse at 80% http://www.badsun.com/images/DV_Samples/Diffusion/3diff_flat.jpg

diffuse map http://www.badsun.com/images/DV_Samples/Diffusion/1diffmap1.jpg

render with diffuse map http://www.badsun.com/images/DV_Samples/Diffusion/4diffmap.jpg


It makes the object look much richer and deeper. Having a seperate channel allows you to use procedurals and gradients and adjusting the opacity of the diff map, just makes it much easier. I will have to try just multiplying the diffuse map over the color and see how that goes. ^_^

I hope this post isn't too large

Slime - no, it's to simulate parts of the object not being as 'bright' due to the surface, rather than the lighting. Imagine a perfect painted ball, except for one scuff mark. That spot would be different than the rest of the surface, either darker or lighter (depending on the paint). To represent it properly, you'd put the scuff in the bump map (to generate the contour) and the diffusion map (to make that spot brighter or darker). You could also brighten or darken that spot in the color map to create the different diffusion level, but that would make it harder to change the shape of the scuff later.

Rumors - Don't get me wrong, I agree that a diffusion map is a very powerful tool. I'm just saying that baking the diffusion map into the color map in Photoshop would give the same effect as using a seperate map in the 3D app.

Using the map in Lightwave is the way to go if you're using that app, since it lets you tweak it without crossing between apps, but you don't get a penalty (other than tedium) to using an app that doesn't support diffusion maps. You just have to do the extra steps of re-combining the diffusion map with the color map in Photoshop every time you want to change it.

Bump maps and specularity maps have to be seperate in the 3D app, since the incident angle of the light is key in the calculations. A diffusion map is angle-indifferent, so you can just combine it with the color map.

[This message has been edited by Das (edited 07-10-2002).]

Unless you use a gradient on the diffuse channel based on the inciednt or light angle ^_^ you can't do that in photoshop.

Later today I'll do some more tests though. I still think that the photoshoped diffuse will still reflect 100 percent of the light in the darker areas, while the diffuse mapped will scatter more for the darker areas.

The Photoshop diffuse can't reflect 100% unless the diffusion map was pure white at that point (in which case the Lightwave diffuse map would also reflect 100% at that point). If Mauritus' math is correct (and it looks good to me), you should get identical results. Lightwave multiplies the maps at render time, ignoring incident angle, and Photoshop multiplies the maps when you choose the multiply blend mode.

quote:

---

Unless you use a gradient on the diffuse channel based on the inciednt or light angle ^_^ you can't do that in photoshop

---

Well, yes, but we were talking realism here. Channels can be 'misused' for all sorts of artistic effects

I think we have a bit of a miscommunication here. What I meant by reflecting 100% of the light. I meant without the diffuse image/channel, the light will bounce off the object 100%. it will not be scattered, so you are getting 100% of the light reflected back. It doesn't matter how light or dark parts of the color map are, it will still bounce back all the light. Maybe this will help explain what I mean.



Diffusion will scatter the cg light so not all of it will be thrown back into the camera.

Incident angle of the diffuse isn't misused at all. I use it all the time for skin and metal and cloth. It's they way things work in the real world. Here is a shirt with an incident angle gradient in the diffusion channel.



Despite this thread changing focus, I'm still sticking to my guns ^_^. That is a color map of the earth, not a diffuse.
(hope the images work)

Ooooh, I see what you're saying. So when the diffuse value is one, it's just like a regular colored object, but when it's zero, it's like a black object with a specular reflection of the color at that point?

Actually, that makes sense. And I can see the usefulness of it.

As I originally stated, it is also a 'diffuse color' map. Just calling it a color map could be considered too unspecific. What kind of color? Specular color? Diffuse color? Luminescence color? It is a fairly common industry tendency to shorten the term 'diffuse color' to 'diffuse' when talking about maps. Diffusion maps and diffuse color maps are not the same thing, obviously, but it's no more wrong to call it a diffuse map than to call it a color map. Both are imprecise, but commonly understood. All imo, of course.

If you go around calling a diffusion map a 'diffuse map', I think you're more likely to be misunderstood, since a lot of people are used to hearing 'diffuse color map' shortened to 'diffuse map'.

Your diagrams are misleading, I think. You're showing pure reflection vs. pure diffuse scattering. The diffusion map doesn't control 'how scattered' the light is. Diffuse light is evenly scattered in 3D rendering. From the article you linked: so basically it's the surface attribute that determines how much light is reflected, as well as absorbed, by the surface (obviously, she's talking about diffuse reflection here). This translates to: high diffusion = lots of light reflected, low diffusion = lots of light absorbed. This directly translates to the perceived brightness of the surface.

You seem to be convinced that the diffusion map controls 'how scattered' the light is at a point. That isn't the case, as Leigh and other posters in the thread you linked state.

Using the diffusion map to simulate fuzzy cloth by driving it off of the incident light is a cool trick, but it isn't simulating the actual diffusion of a real surface. That's what I mean by misusing a channel for artistic effect. You can do some very cool things (and some very wierd things) by using channels in ways for which they were never intended.

Metals are normally simulated by using the reflection channel (so they reflect the environment as well as the incoming light), and a colored specular highlight (in the case of colored metals). Diffuse color is virtually nil on shiny metal, so the diffusion map would have very little effect.

No, Slime, that's what I mean by the diagrams being misleading.

If the diffuse value is 1, the color is exactly the diffuse color of the object at that point.
If the diffuse value is 0, the color is black, regardless of the diffuse color at that point (100% of all light will be absorbed).

After the above calculation, the lighting is taken into account.

Ok, I couldn't find my Lightwave 7 manuals, but I found the ones for version 6.

Page 9.7:
Diffuse (sometimes called diffusion) is the amount of light scattered by a surface. A high level scatters a lot of light and, therefore, the surface will appear bright. A low level absorbs most of the light and, therefore, the surface will appear dark and dull. Metal and dirt surfaces are good candidates for a low diffuse level. Common values are 40%-80%. Surfaces need to have some diffussion in order for shadows cast on them to be visible. (The misspelling of diffusion in the last line is theirs, not mine)

Note that they don't say it controls 'how scattered' the light is, but they do explain that high diffusion yields bright surfaces and low diffusion yields dark surfaces. This is what I've been saying.

I see why we're having the argument over terms. Everywhere I've seen, the full name of the color channel is 'diffuse color', and the diffusion map is a separate entity. Lightwave uses 'color' for what most people call 'diffuse color' or just 'diffuse', and they use 'diffuse' for what most people call 'diffusion' or 'diffusion map'.

Welcome to the wonderful world of non-standardized-CGI-terms. There are a lot of them.

Might as well make it four in a row (assuming no one sneaks a post in).

I'd like to stress that I'm just arguing my case. I'm not trying to flame you here. I know I can come across harsher than I intend when I get to arguing a point.

Just don't want to scare off someone new (just noticed your post count)

I see what you mean and I was wrong. I had a look at some maya sites and they do use diffuse color, diffuse spec, etc. I've always heard the term diffuse and it only meant one thing.

Semantics are a bitch.

But you can control how the diffusion scatters the light because it's based on the diffusion map you make for it. If you want an area dark, paint it dark and it will absorb, want it bright? paint it white and it'll scatter. Maybe you are refering to a different level of control, but that's how I read your post.

Oooooooh.

And I just realized that POV-Ray *does* sort of support this, indirectly:

#declare diffusecolor = pigment{rgb <1,0,0>}; // happens to be all red
#declare diffusionamount = function {f_noise3d(x,y,z)*.4+.4}; // various shades of gray
#declare finalpigment = pigment{function{diffusionamount(x,y,z)} pigment_map{[0 rgb 0][1 diffusecolor]}};

Maybe we're just arguing terms. I'm saying the diffusion map controls how much light is diffusely scattered, but not how scattered. That is, the light isn't narrowly scattered or widely scattered because of the level of the diffusion map. It's always scattered all over the place. (Technically, it isn't actually scattered at all, the renderer just predicts how much light would reach the camera if the lights scatters every which way from each point on the surface).

What you seemed to be saying (and I may have read you wrong) was that the diffusion map somehow controlled the direction of the scattering, or how narrowly it was scattered, or some such.

Wait, let me scroll back:

quote:

---

What I meant by reflecting 100% of the light. I meant without the diffuse image/channel, the light will bounce off the object 100%. it will not be scattered, so you are getting 100% of the light reflected back. It doesn't matter how light or dark parts of the color map are, it will still bounce back all the light. Maybe this will help explain what I mean

---

There, that's what I was objecting to. The light is scattered no matter what's in the diffuse channel (as long as it isn't solid black - that would cause all the light to be absorbed). The color channel (using lightwave terms here, so we're on the same page ) describes the diffuse color. Any light reflected because of the color channel is automatically diffuse - autmatically scattered. If you set the diffusion channel to 10% or 100%, the light is still scattered just as much - there just isn't as much of it at 10%.

The balance between the reflection map and the color map determines how scattered the light is. Light reflected due to the reflection map is angularly reflected (like a mirror - as in the top diagram in your post up there). Light reflected due to the color map is scattered (as the lower diagram - in every which direction). The diffusion map just determines how much of the incoming light (after reflections) is allowed back out, as opposed to being absorbed.

Think of it like this:
100% light hits the object
The reflect map is set to 50%, so 50% of the light is angularly reflected (all in one direction)
This leaves 50% of the light hitting the surface.
The diffusion map at that point is 80%, so 80% of that 50% will be scattered by the color map (40% of the original) and 20% of the 50% will be absorbed (10% of the original)

So with reflect at 50% and diffusion at 80%, you have 50% reflect + 40% color + 10% absorbed = 100% of the original light.

Naturally, with Lightwave (or most other 3D apps) you aren't required to account for 100% of the light. You can have it add up to any total, but keeping it near 100% makes for realistic images.

Another key thing to remember is that even if the color map scatters back 40% of the light in the above numbers, it only scatters back the full 40% if the color map is 100% intensity at that point. If you darken the color map, it scatters back less of the incoming light, which is why a dark color map at a given point is the same as a bright color map at a given point with a dark diffusion map at that point. The diffusion map is just much easier to work with than continually painting dark smudges in just the right spots on your color map.

This is my key point. If I have a color map that's 100% red at one point, with a diffusion map that's 80% at that point, it will scatter exactly as much light as if I'd used a color map that's 80% red at that point and a diffusion map that's 100%. In one case there's 100% color and 20% is absorbed to leave 80% color scattered, and in the other there's only 80% color but it's all scattered.

[This message has been edited by Das (edited 07-11-2002).]