High Definition Props and Textures - Discussion thread

[Orange_o_]

As I can understand the point which require an explanation is the increase of the model:

With Gmax (I do not know 3dsmax). The preview is more beautiful than the export which is too much pixelated. I thus proceed as it:

1. I export my model normally
2. With Gmax, me resize my model of 200, 400 500 (variable according to the model)
3. I make a preview North South West East with a screen shot
4. On Photoshop I recreate the Apha whith more detail, if I would some more of detail I resize again the pics
5. Before I reduced the image to find the initial size of the 1st export. But now that I know the HD technique , I would reduce until x2
6. I create the FSH and I replace the original FSH


It is necessary to keep in memory that the principle is: the preview is more beautiful than the FSH to export

Orange

[SimFox]

Very good and often used method, of course. I can't figure it too, why Orange first shrinks files down and next enlarges it back (yes, less, but always). Downsampling at such dimensions degrades small details irretrievably, even best interpolation won't restore them. Why not downsample image once, using most suitable Adobe's bicubic resampling? You can obtain your "blending" without losing details.

This is exactly what I was wondering about. He has a large file from the screen grab of the preview (500% normal size). Then he downsize it to what would be normal size (20% of that 500%) and then enlarge again (to 400%).
And I had not idea what he does with it since from picture only action I saw was making the mask...

Warrior:
It's kind of funny ... this discussion we have... Basically we say same thing, only with different title...
I say they are different on inside, you say same , however when we describe this "inside" we basically write the same thing...
I think it all boils down to "liberal" use of terms... I mean you can say that something looking different from different angle because it is a box... just because it's not really not true, it's just irrelevant. There is nothing in specific boxy properties that would even start to explain this. Same like Isometric model is irrelevant term, it's like saying that milk is spacious... Not really untrue because it's not a property milk has...


As per suggestion of bypassing the BAT naming limit by manually adding and naming stuff... Well if you would lay out the plan how to do it I'm willing to try..

Orange:
It's not the point that got me and I believe Enedi wondering...

What we are wondering is:
1. After you have get you "magnified" preview rendered in GMAX (btw as I've mentioned the preview looks better because it uses anti-aliasing which is turned off for the export), you make a screen grab. Right?

2. Then you paste it into Photoshop (into new file there). Right?

3.what do you do right after that? From the tutorial you have posted it looks like you imediatelly reduce it size to when would be original (not magnified) preview render. Is that correct?

4. why don't you use alpha generated by GMAX? Why do you think that alpha you recreate in Photoshop has more detail? Do you manually add it?

[Diggis]

4. why don't you use alpha generated by GMAX? Why do you think that alpha you recreate in Photoshop has more detail? Do you manually add it?

I'm confused about this question.  The alpha generated by GMAX would be for the smaller,  crappier, un AA'd render.  If Orange is using a screen capture from GMAX preview then there is no alpha available and he would have to create his own.

[SimFox]

If Orange is using a screen capture from GMAX preview then there is no alpha available and he would have to create his own.

Why is that?

Alpha is always available!

So if he is doing preview of blown-up model the alpha for this very preview is generate at same very time. and it is right there on the screen! All that it requires to see it just one mouse click!

there!

[Diggis]

Aye? I knew it was with Max, but not in GMAX, thats good news then.    I would assume that orange didn't know about it.

[Orange_o_]

Orange:
It's not the point that got me and I believe Enedi wondering...

What we are wondering is:
1. After you have get you "magnified" preview rendered in GMAX (btw as I've mentioned the preview looks better because it uses anti-aliasing which is turned off for the export), you make a screen grab. Right?

2. Then you paste it into Photoshop (into new file there). Right?

3.what do you do right after that? From the tutorial you have posted it looks like you imediatelly reduce it size to when would be original (not magnified) preview render. Is that correct?

4. why don't you use alpha generated by GMAX? Why do you think that alpha you recreate in Photoshop has more detail? Do you manually add it?

1. right
2. right
3. Effectively I reduced directecment in the size of origin because I did not want to go into the detail of the modification which I could bring it. It's a simply tuto.
4. Most part of time I use the apha generated by Gmax, but sometimes as it is the case for the vegetable, I prefer to make him I even to add details or to adapt. It is subjective
As we can see on the tractor, wheels are more round for example... it is small 

It is the technique which I use recently of time, and it allows me to improve the depiction of prop very cut.
I use it especially to modify quickly Bat that I could not make with Gmax. I would show you an example this evening.

Orange

[SimFox]

yeah especially given that it has text RGB Alpha right there staring you into en eye





you know this is one of the main reason I couldn't understand the whole point of that procedure...


But I need a help now..
I have decided to follow suggestion posted here by few members of simply going for slabbed s3d in the place of single slab one...
And run myself into the impass.

What I did.
1. I exported model as usual.
2. scaled it to 200% (thanx Callagrafx, I've finlally started to use Xref)
3. exported new bigger LODs Exported them and then exported the model itself.
4. deleted from first SC4Modle file all Zoom5 S3d's and FSHs
5. copy/pasted relevant files from second SC4Model
6. fixed Group ID to match target file one.
7. and... still have total mess.

Somehow the Group ID in the material (in s3d) either shows something entirely irrelevant, or if it is correct still slab is simply white - sign of missing FSH, rather then properly textured.
What do I do next?
I tried:
Set Texture
Replace Texture
Change Instance/Name
to no result what so ever.
Remove texture removes it alright, but None of the command seem able to place anything back!

[Gwail]

This is exactly what I was wondering about. He has a large file from the screen grab of the preview (500% normal size). Then he downsize it to what would be normal size (20% of that 500%) and then enlarge again (to 400%).
And I had not idea what he does with it since from picture only action I saw was making the mask...

You have answered our question, SimFox, I've just figured it out. Orange downscales to 20% one file (for texture) and enlarges another copy for making own alpha. Take a look at the calculated file sizes at the top of PSD dialog boxes in both cases.

Somehow the Group ID in the material (in s3d) either shows something entirely irrelevant, or if it is correct still slab is simply white - sign of missing FSH, rather then properly textured.
What do I do next?

You're doing this with Reader, right? Have you tried to reload file? Sometimes Reader still refers to the deleted fsh, loaded previously into the memory at opening the file. Or crashes, when the fsh has been replaced by another one.

[SimFox]

Thanx a bunch! that did the trick!

[Diggis]

yeah especially given that it has text RGB Alpha right there staring you into en eye

Never stared me in the eye mate... haven't done preview renders in gmax at all.    Makes sense though, just I knew the rendering in GMAX was limited so didn't know the preview had those options.

Looking good though guys.  When using Max, would it not be easy enough to do a render from the default cameras using a higher resolution to cut the images up?

[buddybud]

Well since i caused such confusion with my quote let me explain what i was under the impression Orange_o_ was doing and what i was trying to explain.

In any illustration you can have multiple approaches. The ones that concern me for sc4d are technical and photo realistic (as much as possible). Technical drawings depend heavily on crisp clear line weight that clearly delineates one surface from another. This is a well defined approach and makes things quite easy to understand. However in reality there actually very few crisp clean lines. In order to get something realistic you have to have a more believable chaotic line system.... especially for things like flora. So in order to get something not look like clear crisp and painted on, you can simply draw at a higher resolution and add all the detail you want and then reduce it.

Sounds crazy but i'm telling ya it works.

Bud

Edit think of it as producing a surface with slight color variations as opposed to a solid color surface.....same sort of thing!!

[jeronij]

Looking good though guys.  When using Max, would it not be easy enough to do a render from the default cameras using a higher resolution to cut the images up?

Finally something that starts to sound good    ¡¡¡ ... SF... what do you think about this way for one slab models??

[Jonathan]

I dunno,
I'm going to stop posting here now

SimFox, if the IID is right and it still shows white in the reader, then you need to save the dat, close it and reopen it.
That is how I do this.
But reindexing and then rebuilding the DIR file might work as well.

Jonathan

[SimFox]

OK
I have a good news to report!

Slabbing things down DOES work! just as Ripple predicted... I guess by now he has to be awarded the title or Oracle! Since he is NEVER wrong.
Given that I, think I know how to implement it in the export scripts so that we would effectively have a Zoom 6 render in stead of zoom 5. Now that I had tested it on my new GLR stops and seen that it works (although manually editing every vertex was a pain in you know what...) I'll certainly try

Warrior:
re-indexing and rebuilding didn't work, but closing and re-opening, luckily, did

[Orange_o_]

If I understood well (sorry) Simfox, you want to replace the FSH and the S3D...
Why the S3D?

[SimFox]

because it is S3d's what you actually see in game... FSHs are just textures applied to them...

I'm right now running test export. It it will be successful, we'll have new fully automated HiDef exporter already tonight!

[RippleJet]

(although manually editing every vertex was a pain in you know what...)

You can always copy the vertices from the S3D file into an Excel spreadsheet.
Then perform the division by 2 in there and copy the new values back from Excel to the S3D file.

Check Chrisim's Tutorial: MODding rendered SC4Models for more information.

That won't of course be needed, once you get that done within the BAT script!

[Orange_o_]

because it is S3d's what you actually see in game... FSHs are just textures applied to them...

But with the 1st export you already have the S3D it is just enough to replace the FSH

[RippleJet]

But with the 1st export you already have the S3D it is just enough to replace the FSH

When doubling the size of the bat, you're effectively making an export for zoom 6.
After that you reduce the size of the model (the LOD box) by halving the vertices in the S3D file.
The textures will remain unchanged though, at 256×256 pixels.

Thus, the FSH textures are now mapped onto a smaller model.
As SimFox showed with the subway prop above, the result can be both better and worse,
compared to the zoom 5 rendered in an ordinary way.
Normally the result seems to be at least on the same level though.

The advantage comes when looking at zoom 6 in the game. In that case, the zoom 5 model is simply doubled in size (both horisontally and vertically), which means the 256×256 textures would be mapped on a much larger area (512×512 pixels, 4 times in number of pixels), increasing the dot size on the screen (to be 2×2 pixels instead of 1×1).

By doing HD props this way, we already have FSH files that are made for zoom 6 though,
and the dot size will remain the same (1×1 pixel) as in zoom 5.

[SimFox]

You can always copy the vertices from the S3D file into an Excel spreadsheet.
Then perform the division by 2 in there and copy the new values back from Excel to the S3D file.

Check Chrisim's Tutorial: MODding rendered SC4Models for more information.

That won't of course be needed, once you get that done within the BAT script!

Yeah I should have...
But now there is no need anymore.
All be totally automatic, no need for any manipulation!
No need to scale model, no need to copy/paste anything, adjust any numbers, etc.
Just press Render BAT and all shall be done for you!

Orange:
what firt export are you talking?
The thing is s3d modles are slies into slabs given camera point of view so as to fit 256x256pix textures. If your want to have hi-def textures/FSH (basically double the size of what suppose to be you would need to adjust not only FSHs but also slabs.