Jump to content

New Texture Compression Formats

Josh

387 views

The Vulkan renderer now supports new texture compression formats that can be loaded from DDS files. I've updated the DDS loader to support newer versions of the format with new features.

BC5 is a format ATI invented (originally called ATI2 or 3Dc) which is a two-channel compressed format specifically designed for storing normal maps. This gives you better quality normals than what DXT compression (even with the DXT5n swizzle hack) can provide.

BC7 is interesting because it uses the same size as DXT5 images but provides much higher quality results. The compression algorithm is also very long, sometimes taking ten minutes to compress a single texture!  Intel claims to have a fast-ish compressor for it but I have not tried it yet. Protip: You can open DDS files in newer versions of Visual Studio and select the compression format there.

Here is a grayscale gradient showing uncompressed, DXT5, BC7 UNORM, and BC7 SNORM formats. You can see BC7 UNORM and SNORM have much less artifacts than DXT, but is not quite the same as the original image.

uncompressed.thumb.png.9759cc5f5afc7c94947bad97a914cdee.png

dxt1.thumb.png.af7bf97f7c1cf6eaaf5b239cfa3577db.png

dxt5.thumb.png.0a450dcd60ac364d839e4c5ee0ed1d83.png

bc7_unorm.thumb.png.d3c5339ff07e4c68477070c854019c12.png

bc7_snorm.thumb.png.c9f5006d09352e3eba2d890bb1911100.png

The original image is 256 x 256, giving the following file sizes:

  • Uncompressed: 341 KB
  • DXT1: 42.8 KB (12.6% compression)
  • DXT5, BC7: 85.5 KB (25% compression)

I was curious what would happen if I zipped up some of the files, although this is only a minor concern. I guess that BC7 would not work with ZIP compression as well, since it is a more complicated algorithm.

  • DXT5: 16.6 KB
  • BC7 UNORM: 34 KB
  • BC7 SNORM: 42.4 KB

Based on the results above, I would probably still use uncompressed images for skyboxes and gradients, but anything else can benefit from this format. DXT compression looks like a blocky green mess by comparison.

I was curious to see how much of a difference the BC5 format made for normal maps so I made some similar renders for normals. You can see below that the benefits for normal maps are even more extreme. The BC5 compressed image is indistinguishable from the original while the DXT5n image has clear artifacts.

norm_uncompressed.thumb.png.641ec4ad732fe481db4cd58e4ae35499.png

norm_dxt5n.thumb.png.251b778e5633a48c0a75aa70da98c7f8.png

norm_bc5_unorm.thumb.png.e6d267d8e39dc97a09c7b806123f2ec9.png

In conclusion, these new formats in the Vulkan renderer, when used properly, will provide compression without visible artifacts.

  • Like 4


3 Comments


Recommended Comments

I replaced our own texture format constants like TEXTURE_RGBA with the Vulkan texture constants and added support for pretty much all the DDS formats that correspond to Vulkan formats.

  • Like 1

Share this comment


Link to comment

Also got Intel’s compressor working, it will make a good converter.

  • Like 2

Share this comment


Link to comment

Here's a nice shot showing the quality BC5 brings to normal maps:

Image1.png

  • Like 1

Share this comment


Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Add a comment...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Blog Entries

    • By jen in jen's Blog 3
      I thought I would share my experience on this; if you're working on Multiplayer, you will need to protect your packets. The solution is simple, let's go through how we can achieve this by implementing what Valve calls "challenge codes". (Some reading on the topic from Valve here: https://developer.valvesoftware.com/wiki/Master_Server_Query_Protocol#Challenge_response).
      Disclaimer: this doesn't cover other security techniques like authoritative server or encryption.
      So, I've worked on Border Recon last year (I think) and I needed a way to protect my server/client packets. There was no need for me to re-invent the wheel, I just had to copy what Valve has had for a  long time - challenge  codes.
      The idea behind challenge codes is similar to Captcha, but not exactly. Think of it like this: for every packet submitted to the server, it must be verified - how? By requiring the client to solve challenges our server provides.
      To implement this we need to have the following:
      A randomised formula in the server i.e.: a = b * c / d + e or a = b / c + d - e, be creative - it can be any combination of basic arithmetic or some fancy logic you like and can be however long as you want - do consider that the longer the formula, the more work your server has to do to make the computation.  Copy the same formula to the client. A random number generator.  So the idea here is:
      (Server) Generate a random number (see 3 above) of which the result would become the challenge code, (Server) run it through our formula and record the result. (Client) And then, we hand over the challenge code to the client to solve (an authentic client would have the same formula implemented in its program as we have on the server). For every packet received from the player, a new challenge code is created (and the player is notified of this change by the server in response). For every other packet, a new challenge code is created. (Client) Every packet sent to the server by the client must have a challenge code and its answer embedded.  (Server receives the packet) Run the challenge code again to our formula and compare the result to the answer embedded on the client's packet. (Server) If the answers are different, reject the packet, no changes to the player's state. The advantage(s) of this strategy in terms of achieving the protection we need to secure our server:
      - For every packet sent, new challenge code is created. Typically, game clients (especially FPS) will update its state in a matter of ms so even if a cheater is successful at sniffing the answer to a challenge code it would be invalidated almost instantaneously. 
      - Lightweight solution. No encryption needed. 
      Disadvantage(s):
      - The formula to answering the challenge code is embedded to the client, a cheater can de-compile the client and uncover the formula. Luckily, we have other anti-cheat solutions for that; you can implement another anti-cheat solution i.e. checking file checksums to verify the integrity of your game files and more (there are third-party anti cheat solutions out there that you can use to protect your game files).
       
       
       
    • By Josh in Josh's Dev Blog 4
      New commands in Turbo Engine will add better support for multiple monitors. The new Display class lets you iterate through all your monitors:
      for (int n = 0; n < CountDisplays(); ++n) { auto display = GetDisplay(n); Print(display->GetPosition()); //monitor XY coordinates Print(display->GetSize()); //monitor size Print(display->GetScale()); //DPI scaling } The CreateWindow() function now takes a parameter for the monitor to create the window on / relative to.
      auto display = GetDisplay(0); Vec2 scale = display->GetScale(); auto window = CreateWindow(display, "My Game", 0, 0, 1280.0 * scale.x, 720.0 * scale.y, WINDOW_TITLEBAR | WINDOW_RESIZABLE); The WINDOW_CENTER style can be used to center the window on one display.
      You can use GetDisplay(DISPLAY_PRIMARY) to retrieve the main display. This will be the same as GetDisplay(0) on systems with only one monitor.
    • By Josh in Josh's Dev Blog 1
      A huge update is available for Turbo Engine Beta.
      Hardware tessellation adds geometric detail to your models and smooths out sharp corners. The new terrain system is live, with fast performance, displacement, and support for up to 255 material layers. Plugins are now working, with sample code for loading MD3 models and VTF textures. Shader families eliminate the need to specify a lot of different shaders in a material file. Support for multiple monitors and better control of DPI scaling. Notes:
      Terrain currently has cracks between LOD stages, as I have not yet decided how I want to handle this. Tessellation has some "shimmering" effects at some resolutions. Terrain may display a wire grid on parts. Directional lights are supported but cast no shadows. Tested in Nvidia and AMD, did not test on Intel. Subscribers can get the latest beta in the private forum here.

       
       
×
×
  • Create New...