Advertisement Jump to content
  • Advertisement


  • Content Count

  • Joined

  • Last visited

Community Reputation

138 Neutral

About agaudreau

  • Rank
  1. The short answer is, you can create a texture in memory and draw to it dynamically. As far as helping you with a "better approach to the problem" it might be easier if you explained some of what it is you are drawing. Are you actually making a 2D mini map? or is it something else?
  2. agaudreau

    Play video using DX 9.0 in fullscreen

    I find that if you get the frame data with direct show you can swap it into a texture and then render that on anything you want. The hard part is getting to the frame data everything else is easy especially if you have written a graphics engine in Direct3D.
  3. agaudreau

    Animating in DirectX 10

    I am going to assume you mean 3D animation. For 3D animation there are a couple things you need to know: 1) Animation is just the interpolation of key frames, that is the same for 2D animation. For skinned 3D animation its the interpolation between bone matrices. 2) Take the interpolated data and apply it to the object being animated. Using these two ideas (and a little research) you can create your own animation engine with any graphics api. By the way, DirectX 10 has no fixed function pipeline built into it anymore so you will most likely be working with shaders for rendering. The vertex shader is the best place to apply the interpolated matrix values to the model's vertices because the GPU was built for speed using vectors and matrices (read float mathematics).
  4. agaudreau

    Customizable Characters

    If a model hiarchy has the same bones as another model with animation. You can apply the other model's animation bone transforms to your model's bones.
  5. Your event system only has one "flaw" (as far as I can see) which is the use of strings. String compare will always be slower than int compare, and in a system that potentially compares hundreds of times per frame 30-60+ frames per second you could be taking a hit on performance. I would use enums defined in a commons header file or something. 1 for event names and 1 for systems. using Enums would save you space and run faster which is always good. Other than that it looks good and I wouldnt worry about "(wrong name used)" because, strings or not it can still happen (ie sending out of bounds number to reference an event that isnt there).
  6. agaudreau

    [dx10]shader ,texture and lights

    Do you mean light baking?
  7. agaudreau

    Object import

    I dont know what format you should use but I think if you are using OpenGL dont use .x files because it would break one reason why you would want to use OpenGL in the first place (multiplatform).
  8. Is the image you are using for your sprite a (for some reason my brain just farted and I cant remember the term but) resolution of: 8x8 or 16x16 or 32x32 or 64x64 or 128x128 or 256x256 or 512x512 or 1024x1024 etc? if its not (enter term I forget here) the video card drivers will auto stretch your image which will do two things: 1: Make it look like crap. 2: Slow down rendering alot because it doesnt just do it once, but every time it needs to render it. -TonyG
  9. agaudreau

    Multitexturing Method Check

    Are you using the programable pipeline or standard?
  10. agaudreau

    Free Sprite Sheets

    I think that people usually create an animation editor so they can load any sprite sheet and define multiple things about each sheet because they are all different. This type of editor would allow you to set anchor points, define the view rect (clip region) and also things like collision rects etc. Putting time into an editor like this can increase content generation greatly. But if your only looking for pre formated sprite sheets that are all the same size: I googled sprite sheets and was the first link.
  11. As my knowledge of threading continues to evolve and grow, so do my questions on how to properly implement them. So now I am at the point where I am going to design my engine for multiple threads and I want to implement them as worker threads the question comes up: "What is a task." But not in the sence that I dont know what a task literaly is, but more what do I make into a task? I have been pondering about it for a while and can only come up with so many. So here I am asking for some creative help: What things, that are not game specific (ie engine stuff) should I be making into tasks? My idea is that tasks can either be syncable (meaning that they run in parrallel until they are done and then wait for the main program to be ready before ending) or they are autonomous (Something that doesnt need to sync, like playing the current frame's sounds). Im still learning as I go but I really want to get this as close to "right" as possible to minimize refactoring/redesigning later.
  12. agaudreau


    This might be the long way around but, if you have the position of all verts including the one you just found, and the UVs of all but the one you just found. Then I believe you can convert them into 3D vectors by adding 1 in the 3rd place and do some vector math to find the % distance difference of the unknown to your longest edge then use that to scale the UV of the biggest UV along that longest edge to get the UV your looking for. I am not sure this will work but it might get the ideas flowing.
  13. One thing to remember is that when hardware companies are doing their benchmarks for tri counts, they usually run degenerate triangles so its not a very good measure to hold up to your program.
  14. agaudreau

    Threaded memory use

    This doesn't require sync, only a way to determine when all handlers for current frame have completed, so you know where to flip.[\quote] Are you sure? I dont want to flip the memory "buffer" if the rendering is not done or if the other activities are not done either. So I would have to wait for both threads to be complete before swapping (syncing yes?). Otherwise the rendering would be not always rendering the same "frame"'s data right?
  15. agaudreau

    Threaded memory use

    Thanks for the reply, this problem has been the bane of this project for a while now. I dont like the idea of having twice as much data for objects but I guess there isn't a good way around it if I want to implement threads the way I want.
  • Advertisement

Important Information

By using, you agree to our community Guidelines, Terms of Use, and Privacy Policy. is your game development community. Create an account for your GameDev Portfolio and participate in the largest developer community in the games industry.

Sign me up!