Hacky Normal Map Correction

Gannon

I was baking out some normal maps for some stones I made for a scene and discovered a nifty little trick. I'm sure it doesn't fix everything but hopefully it helps some people in a pinch.



Sometimes I just can't get a good bake on objects so I'm glad I stumbled upon this. I'd love to see how this can be used on other models and situations.

Ace-Angel

I don't understand, you're killing the gradient, which with a proper smooth-group setup should give the same results?

Unless I missed something, because it looks like your Normals are one group as opposed to split by UV cuts.

Gannon

Maya doesn't use Smoothing groups per say, It's just defining hard and soft edges. The closest way to mimic them would be going through and hardening the outside edges of the shell/UV island and softening the interior ones.

passerby

@Gannon the same thing still applies to getting a clean normal map bake, if you would like i can post a mel script i use that automatcally hardens the edges of UV islands for you?

Gannon

I know, this just a hacky trick is all.

A script would be fantastic. I've been doing it manually for a while and sometimes that's pretty tedious.

passerby

just toss this on a shelf, and run it on a object, after you do your UV's

string $objList[] = `ls -sl -o`;
string $uvBorder[];
string $edgeUVs[];
string $finalBorder[];

for ($subObj in $objList) {
select -r $subObj;
polyNormalPerVertex -ufn true;
polySoftEdge -a 180 -ch 1 $subObj;
select -r $subObj.map["*"];

polySelectBorderShell 1;

$uvBorder = `polyListComponentConversion -te -in`;
$uvBorder = `ls -fl $uvBorder`;

clear( $finalBorder );

for( $curEdge in $uvBorder ) {
$edgeUVs = `polyListComponentConversion -tuv $curEdge`;
$edgeUVs = `ls -fl $edgeUVs`;

if( size( $edgeUVs ) > 2 ) {
$finalBorder[ size( $finalBorder ) ] = $curEdge;
}
}

polySoftEdge -a 0 -ch 1 $finalBorder;
}

select -r $objList;

Gannon

sweet, Thanks!

unstoppablex

pretty nice trick for removing unwanted normal map shading!

thanks for sharing this

Sandro

Okay, nice hack but you might as well cut out whole baking process and generate your normal from diffuse. Because resulting map doesn't have any big or medium surface definition left, just surface dents and noise.

r_fletch_r

why not just fix your models smoothing? like sandro says, your totally loosing the advantage of baking a high to low normal map.

Bal

This process doesn't make any sense at all.
If the gradients are there after your bake, it's that they are supposed to be there to compensate for the lowpoly shading, or you did something wrong. You shouldn't have to hack around in Photoshop like this.

And your final image still looks wrong, there are some weird inner bevel type effects on all the edges.

WarrenM

This has come up in other normal map threads but yeah, what they said above me. If you have to hand tweak the normal map afterwards, that means you're going to have to do that every time you update the normal map. That will add up to a lot of busy work over the course of a project. A LOT of busy work.

Better to fix your model and get a clean bake.

Gannon

@ sandro - Baking a diffuse to normal isn't going to give you the right results at all. The information is usually not what you want most of the time. You usually want a height map but then you lose out on edge quality so you compensate through extra geo on edges.

Yeah, just changing the smoothing groups can help but it doesn't always. Sometimes it come down to the geo, so without retopoing the entire thing this is just a quick hacky fix.

EarthQuake

oh god, terrible terrible idea, just fix your model!

[HP]

EarthQuake wrote: »

oh god, terrible terrible idea, just fix your model!

This!

You can't polish a turd.

Gannon

rofl, fiiiiiiiiine.

[HP]

Btw Gannon, we (I guess I can speak for us all here) are not dissing on you, or the fact that you created this thread.

You went ahead and created a thread, even wrote a guide, and show people this. That's more than 95% of us ever do for this community, so we all learned here, not only you!

Don't feel reluctant next time you wanna share possible new discoveries of workflows, etc.

r_fletch_r

[HP] wrote: »

Btw Gannon, we (I guess I can speak for us all here) are not dissing on you, or the fact that you created this thread.

You went ahead and created a thread, even wrote a guide, and show people this. That's more than 95% of us ever do for this community, so we all learned here, not only you!

Don't feel reluctant next time you wanna share possible new discoveries of workflows, etc.

^ This.
Hopefully we werent over zealous. There just so much of this stuff going about that it gets frustrating.

EarthQuake

Right, the biggest problem with this sort of things is they have this horrible tendency to snowball. You post some seemingly harmless bit info about something you've done, and all of a sudden we'll see countless people who don't really understand what or why you've done it repeating it because they think "gradients in normal maps are bad". Even worse, these people start popping in to other threads in tech talk and telling them to follow the advice in this thread, instead of taking the time to learn and understand WHY smoothing errors occur in the first place.

Its a constant battle of miss-information.

unstoppablex

I don't know why everyone is making a huge issue. It's a good trick if you do have unwanted normal map shading. Of course you should always make sure you have the best possible bake but we all know normal maps are not always perfect, and this could be useful at some point.

i think the only problem with this thread is the title and it's approach. like many of your mention, this is the intention of normals from the bake and this isn't a general problem.

r_fletch_r

unstoppablex wrote: »

I don't know why everyone is making a huge issue. It's a good trick if you do have unwanted normal map shading. Of course you should always make sure you have the best possible bake but we all know normal maps are not always perfect, and this could be useful at some point.

Its not a good trick, its a hack that tries to make up for incorrect execution.
If you model the base correctly you wont get this issue. No one should ever recommend visually editing normal map gradients.

take a little look at Earthquakes work and you will see what can be achieved with correct normal map workflow.

lluc

Here my technique, may be it isn't perfect !!

cryrid

Poor gradients, so misunderstood : /

lluc

is it so bad? How can I improve? I'm very interested ! pls!

Ace-Angel

cryrid wrote: »

Poor gradients, so misunderstood : /

Gradient are the tool of the Witch! Burn the unclean edges! Melt the abomination of the baker!

zakhar2

Mods please rename this thread "Bad Normal Map Tips ITT", thx

m4dcow

zakhar2 wrote: »

Mods please rename this thread "Bad Normal Map Tips ITT", thx

Or parse the thread title for "Normal Map", and have a confirmation box verifying that the poster has read EQ's normal map thread.

Computron

m4dcow wrote: »

Or parse the thread title for "Normal Map", and have a confirmation box verifying that the poster has read EQ's normal map thread.

Lol.

Good ol' Earthquake & 3-point Folk.

EQ, you really should think about putting some that Polycount Teaching money toward a gold chain. (They are paying you, right?) :poly124: :thumbup:

Froyok

Hahahaha !
Computron, this is awesome ! x]

joebount

A normal map is meant to describe the difference between a highpoly and a lowpoly, it's NOT a bumpmap.

r_fletch_r

does anyone have a good link that describes why this is wrong. I think we've told the OP they are wrong enough times now

Computron

It was already posted, the stickied normal map waviness thread.

cptSwing

cryrid wrote: »

Poor gradients, so misunderstood : /

i lol'd!