textures, specular, dds, compressions and more: clarifications

[eliasfsx]

hello everyone...
I have several questions and some results to share...(also this will be useful for those who also have the same doubts or are looking for solutions)
To summarize, it's about the "SPEC" file, and DDS, and compressions (and differences).

 

SPEC files
I never understood what they are for... under what conditions can someone say "oh!! I see it"? (day, evening, dusk... with a certain type of lighting... if the paint is matte or glossy... etc) Does it affect anything (visually or quality) if the resolution of the spec file is reduced?

Right now I'm going through two different cases of textures...and I wanted to take advantage of the answers....
case 1: the first plane has a GREAT alpha layer... (the model is "polished metal" without colors, just metal color, VERY shiny)... how does it affect that the SPEC file is in low resolution? (or compressed in dx1)
case 2: the same model but with a livery with almost no shine (alpha almost nonexistent)... the livery implies that the model is "worn" and "dirty" matte color....how does it affect the file Is SPEC in low resolution? (or compressed in dx1)
(until now I did not modify these files at all)

 

I want to clarify... that the official paintkit of the model (and the creator of the model) advises saving the specs in dds dx1... but I also wanted to reduce the size by half (very small)... to obtain maximum performance (. .if the visual difference is not noticeable)

 

update 3: I forgot to mention that I have other models that DO NOT EVEN have spec files (another model)..of course if it comes with it, I keep it,... but if it doesn't come with it, why doesn't it need it?

[eliasfsx]

The other issue (although this is not important) is how to edit a "cubemap" (env map...globalenv....chrome...or whatever they are called)... they are those "cross" shaped files that contain a reflection of the sky and ground (may vary)...whenever I try to edit it..the model cannot read it..and the model ends up without reflections (even though it has the full alpha)...this only happens when I edit the cubemap.

[eliasfsx]

The other topic is DDS FILES and compression
(sorry for the long introduction... if you want, skip it... at the end below, I have the question)

I understand that fsx handles 2 types of files: bmp and DDS... and to detail, DDS handles different compressions and qualities for different uses, the best known are:

DDS DX1 (with image compression... there are also 2 versions... with or without alpha)
DDS DX5 (with image compression... almost no one uses dx3... because dx5 is equal and better) (with alpha channel)
DDS 888-8 32 bits (sometimes called "extended") this format DOES NOT HAVE ANY COMPRESSION... everything you edit will be saved without "degradation" (it will be "pure") (includes alpha channel)

There are advantages and disadvantages to each of them...
the DX 1 and DX5 versions you must handle it as a "single shot"... try to do all the work in a single attempt... since the result will be degraded when saving... and the worst thing is... the more you open you edit and If you save the file in "DX", the more degraded it will be (ugly violet or colored squares will appear, and the natural color you painted will begin to change to a similar one)
..the advantage is that it is much lighter than other save formats.

the DDS 888-8 32 bits the advantage is that you can edit it as many times as you want... it will NEVER lose its quality (as long as you save it as 888-8 32 bits)
..the disadvantage is that it weighs FOUR TIMES MORE than the DX versions (depending on the model...and the number of textures....it could be a RAM memory problem in fsx)

someone recommended me here on flightsim "for your editing work and projects...save it in 888-8 32 bits....but when you want to publish the work...publish the copy in dx5" (that is, keep a backup from the original without any degradation... but published in dx5)

I load the texture in DXTBMP as a program to manage dds (there the texture and the alpha are displayed separately)... I edit with PAINT.NET and save.... then I "refresh" dxtbmp to show the change and save with dxtbmp ...in dx5 or 888 (as you wish)

but ARE DIFFERENT TYPES OF COMPRESSION (OR SAVE) in DX5 better than others?
(Are there other programs that can save in DX5...better than in DXTBMP?)

(I am not disparaging the glorious work of Martin Wright and his dxtbmp... rather seeing that many years have passed since his creation... I was considering knowing if other programs CAN (I am asking) compress in DX5... better ) (that is, with less degradation)

 

why?
I know that a dds MUST be opened with dxtbmp... but as an experiment... I opened a texture DIRECTLY with PAINT.NET... (the "squared grid" was visible... it means that the alpha layer is not separated). ..don't edit anything...just save the file as dx5...
comparing the original (888-8), and the dx5 from dxtbmp and the one created by paint.net...there are differences

(I clarify that the texture was 100% GRAY in color)

both dx5 have that degradation (small purple squares, blurring and slight color changes...product of compression)....but there is a difference between what is created by dxtbmp and paint.net....
in dxtbmp the color tends to change "towards violet" (almost gray, but still violet)... and the version created in paint.net tends to change "towards green" (slight green tone... but more noticeable than the violet by dxtbmp)

The funny thing is that paint.net has compression options for dx5... different types of "recognition" to avoid stains... even so, it changes more than what is done by dxtbmp.

So the big question is: ARE DIFFERENT TYPES OF COMPRESSION (OR SAVE) in DX5 better than others? (Are there other programs that can save in DX5...better than in DXTBMP?)

[eliasfsx]

Any ideas... or recommendations?

[jgf]

The alpha layer is a 256 color greyscale image, the darker shades are more reflective, lighter shades less reflective.  If you have no alpha layer the plane will have no "shine", just all flat colors;  some aircraft are modeled with no alpha, it cannot be added since that requires the mdl file to be flagged for it.

 

DXT images for FS must be 32bit, but most graphics programs, such as Paint.net, save in 24bit, hence the need for DXTBmp, which saves in 32bit (the extra 8 bits are for the alpha channel, even if one is not used). 

 

For best results never send the image from DXTBmp for editing, this results in multiple processing and loss of detail (as you've seen - color blotches and blurry imaging).  Do all your editing in 24bit bmp format, and always save that master image;  when done make a copy and open that in DXTBmp and save.  The same with the alpha image, though it isn't as critical;  if you do not have an alpha, create it in DXTBmp and immediately export it and save as 256color bmp, adjust the shades of grey and import to DXTBmp and save the entire image once.  If more edits are needed work on the master images and import them to DXTBmp again.  If you suspect issues with proper DXT format, copy and rename one of the original aircraft texture folders, open the file in DXTBmp and "send to editor" (leave DXTBmp open), copy your image into that file, save and exit, back in DXTBmp use "reload after edit", and save, this will save the file in the same format as the original.

 

A good basic description of DXT formats - http://wiki.polycount.com/wiki/DXT

[eliasfsx]

4 hours ago, jgf said:

The alpha layer is a 256 color greyscale image, the darker shades are more reflective, lighter shades less reflective.  If you have no alpha layer the plane will have no "shine", just all flat colors;  some aircraft are modeled with no alpha, it cannot be added since that requires the mdl file to be flagged for it.

 

DXT images for FS must be 32bit, but most graphics programs, such as Paint.net, save in 24bit, hence the need for DXTBmp, which saves in 32bit (the extra 8 bits are for the alpha channel, even if one is not used). 

 

For best results never send the image from DXTBmp for editing, this results in multiple processing and loss of detail (as you've seen - color blotches and blurry imaging).  Do all your editing in 24bit bmp format, and always save that master image;  when done make a copy and open that in DXTBmp and save.  The same with the alpha image, though it isn't as critical;  if you do not have an alpha, create it in DXTBmp and immediately export it and save as 256color bmp, adjust the shades of grey and import to DXTBmp and save the entire image once.  If more edits are needed work on the master images and import them to DXTBmp again.  If you suspect issues with proper DXT format, copy and rename one of the original aircraft texture files, open the file in DXTBmp and "send to editor" (leave DXTBmp open), copy your image into that file, save and exit, back in DXTBmp use "reload after edit", and save, this will save the file in the same format as the original.

 

A good basic description of DXT formats - http://wiki.polycount.com/wiki/DXT

first of all... thank you very much for responding...
1-what I meant is... the SPEC layer was visible or "noticeable" if you have a VERY bright alpha layer... what do you think?
(this is a question I wanted someone to help me with..)
because my idea is to reduce the size (and the quality of the SPEC, taking it to dx1) to a minimum... for performance reasons.
Is the spec visible or noticeable when there is VERY bright alpha layer? (see image below)

---
about the formats, and the bits... as long as I create, edit and save in 888-8 32 bits... there was never any data loss or "blobs"... this happens ONLY when I convert it to dx5 (or less )
about paint.net... I just confirmed... it can save bmp in 32 bits (in addition to others)... and dds... one of the save options is "B8G8R8A8" (blue, green, red, alpha) (it is the same as the 888 32 bits)..and it does not lose any quality.
What I meant was that dxbmp and paint.net can save in dx5...and that is where the differences are noticeable.

I understand (as you said) using and editing a "master" outside of dxtbmp... and then using dxtbmp to save it in the required format

What I want to add is that... opening the texture from dxtbmp... AND ALSO from paint.net... as long as you save in "888"..or "b8g8r8a8" you will never lose quality or damage the image... ( I've seen this for years)

[eliasfsx]

another matter (nothing to do with the previous one) (please do not confuse)...

I want to download (from the internet) and EDIT a texture... but:
- the texture (downloaded from the internet) is saved in dds dx5
-It is clearly noticeable that there is a "reduction" in quality for having saved like this in dx5
-there is no "master" or anything like that...just the textures (typical)

I would like to edit the texture...but resave in dx5 would stain it even more... any suggestions?

I do not want it to be "restored" to something similar to a "master" (it is impossible if it is not available)
I only want to save in DX5...but without further staining it (going to 888 32bits will preserve the quality...but there will be a considerable increase in weight...)

(the only thing I can think of is to "reedit and clean" the texture...save in 888 32 bits...and then convert a copy to dx5)
what do you think?

[jgf]

If by spec layer you mean the actual texture (image) file you cannot change its formatting - size or color depth -  or it will not display properly (may not even load).  The alpha layer can affect how the texture displays beyond just how reflective it is;  the stronger the alpha channel (darker) the lighter the base textures will appear (it is almost impossible to have a very shiny black aircraft because the alpha will make even pure black look dark steel grey).  So you usually paint an aircraft darker than what you want so it will display correctly once the alpha is applied.

 

I am not conversant with newer versions of paint.net as I stayed with Win7;  most of my editing is done with Paint Shop Pro, it has vector functions and more options.

[jgf]

10 minutes ago, eliasfsx said:

(the only thing I can think of is to "reedit and clean" the texture...save in 888 32 bits...and then convert a copy to dx5)
what do you think?

 

That may work, just make sure the DDS conversion is the very last thing you do.  Try it and let us know, if it doesn't work all you've lost is some time (I hate to think of the time I've wasted ...er, spent ... trying to get a livery to appear as I wanted;  occasionally even scrapping the work and starting over).

 

The main thing is to always have a good clean bmp master file, so if the finished product isn't acceptable you can always go back to that file and try again.

[eliasfsx]

9 hours ago, jgf said:

If by spec layer you mean the actual texture (image) file you cannot change its formatting - size or color depth -  or it will not display properly (may not even load).  The alpha layer can affect how the texture displays beyond just how reflective it is;  the stronger the alpha channel (darker) the lighter the base textures will appear (it is almost impossible to have a very shiny black aircraft because the alpha will make even pure black look dark steel grey).  So you usually paint an aircraft darker than what you want so it will display correctly once the alpha is applied.

 

I am not conversant with newer versions of paint.net as I stayed with Win7;  most of my editing is done with Paint Shop Pro, it has vector functions and more options.

I understand you mate... I also hate thinking about the time wasted on the "failed" edition.
the SPEC file ...is a texture (a dds) almost identical to the visible texture... there is a spec file for each of the textures (in some models)
also called "specular"...if the main (visible) texture is called "fuselage_L_back.dds", the SPEC will be called "fuselage_L_back_spec" (or "_specular")... they are usually more diffuse than the original texture.

(to give an example of which model the spec comes with... Manfred Jahn's Douglas DC3 for fsx has them in each of its liveries)

I suppose it gives some lighting detail... but if the alpha layer is on top... I think it won't matter too much if I reduce the quality or size

[jgf]

Ah, a specular layer is usually highlight and shading;  if you paint an aircraft a solid color it doesn't look right because there's little 3D effect of just one color over the model, so you add highlights to areas that would have more light on them and shadows to areas that should be darker.  This makes painting easier since you can just apply the color(s) and let the shading layer tone the color.  The alpha is applied over all this;  alpha is not a layer as such but is a map telling the program how reflective each area should be.

 

This may help - https://www.sim-outhouse.com/sohforums/showthread.php/74535-Specular-maps-in-FSX