[Bf-committers] ...depth of field
uneboite uneboite
bf-committers@blender.org
Wed, 16 Jul 2003 03:40:53 +0000
<html><div style='background-color:'><DIV>Well, well, it seems that I am the very newby in the optic field.</DIV>
<DIV>And I still have to understand the rendering code in Blender.</DIV>
<DIV>I do have very solid backgrounds in C programming though and I would be very happy</DIV>
<DIV>to make a little contribution for Blender, so I'll try to learn.</DIV>
<DIV>If I don't succed with dof, well I'll ask Blender's gurus how I can help on other</DIV>
<DIV>topics... because I have some time to do it and I have good programing experience...</DIV>
<DIV>So I can be of some help.</DIV>
<DIV> </DIV>
<DIV>Using floating point data is very interresting and should give more precise results, but</DIV>
<DIV>if I use the convolution method (or a method combined with convolution) I don't see yet</DIV>
<DIV>how I can do it (convolve) without the full image already generated in R.rectot. Can somebody give me a hint please? Are you talking about using a full-formula filter, instead</DIV>
<DIV>of convolving with matrices?</DIV>
<DIV> </DIV>
<DIV>I'm not against using other techniques than convolution (although I told the contrary</DIV>
<DIV>in my first posting). The only other technique I've read a bit about (in the OpenGl</DIV>
<DIV>red book I think) is jittering and on first thought I think it would be very slower than convolving.</DIV>
<DIV> </DIV>
<DIV>I'm playing with an idea in my head about an approach similar to convolution but that could permit me to work with alpha channel. In a convolution filter, a pixel "sucks" values from its</DIV>
<DIV>surrounding pixels. I would do it the other way, I would "blow" the current pixel value on the</DIV>
<DIV>surrounding pixels. Combined with a multi depth value-and-Z-buffer it could resolve some problems...</DIV>
<DIV> </DIV>
<DIV>Thanks a lot for the websites about dof and the "fourier optics" hint.</DIV>
<DIV> </DIV>
<DIV>One of my concern about using the F stop approach and trying to simulate the real</DIV>
<DIV>lens behavior with camera-like parameters is that Blender is not working with real life</DIV>
<DIV>units. When I do my calculation involving distances, how do I treat units?</DIV>
<DIV>What 1 means in Blender? 1 inch, 1 millimeter? We're not talking about meters or inches. That's why I was thinking about using objects (among other things) to make the calculations more supples. Although less scientific an approach, I though it could be convenient, let's</DIV>
<DIV>say if one works with different "scales" between his scenes.</DIV>
<DIV> </DIV>
<DIV>Would you please inspire me on this subject?</DIV>
<DIV> </DIV>
<DIV> </DIV>
<DIV>Thanks a lot</DIV>
<DIV> </DIV>
<DIV>Uneboite</DIV>
<DIV> </DIV>
<DIV> </DIV>
<DIV> </DIV>
<DIV> </DIV>
<DIV> </DIV></div><br clear=all><hr>Help STOP SPAM with <a href="http://g.msn.com/8HMKENCA/2731??PS=">the new MSN 8 </a> and get 2 months FREE*</html>