Integrity
06-04-2006, 04:29 AM
I'm a little confused on the placement of hdr balls.
Pretty soon I want to do a simple scene of a dragon landing on a famous landmark where I live. It's an asian pagoda (if that's the right term) atop a small mountain that's part of a historical site, with a road passing by it. The dragon will simply fly in overhead the camera (which will be positioned across the street looking towards the pagoda) and land on the sideroof, and will basically just act angry, roar and knock off a whole bunch of shingles (sort of like what the dragon did that Harry Potter fought in the most recent movie).
Recently I made a (very) short film where I used an HDR ball for lighting with radiosity (as the background); the position of the ball was pretty much where the CG object was, and I positioned the texture in 3D space in exactly the same spot. I also created a simple plane object that used the same HDR probe as a texture that matched the spherical projection of the background lighting, thinking that this was the correct way so the reflected light that came from the ground (which was stored in the probe image I was using) would be in the right position and have the right intensity when lighting up the shadow areas of my CG object. I offsetted the texture by the heighth of the ball in the real world, which automatically textured the plane correctly with the image probe (besides the distortions caused by the spherical mapping); sort of like camera mapping I guess. Should I really have done this? Or would simply using the background lighting have been good enough? Because when I did this it looked better.
The predicted problem that I think I'm going to face now with the dragon is that I will not be able to take photographs of the ball in the same position as the dragon will be (once it lands on the roof, way out of my reach). I was thinking of just taking them where the camera will be positioned, and doing what I did in my previous short film; but my confusion and problem is that instead of offsetting texture mapping only a couple feet (where I had thought the distortions would be negligible), it's going to be probably a hundred feet or more this time. My confusion is that when making light probes from reflective balls does the ball need to be positioned where the CG object will be? Or can you take it anywhere where it's feasible and be able to place your CG object anywhere in the HDR probe lit background and the lighting information be accurate? Shouldn't, let's say the tail end of the dragon, have a different light probe than the head end? Since they're further apart? Is there a specific distance when this technique should be used? The fact that the dragon will be sitting right on the pagoda where the bounced light from it will effect him more than what the light probe of the position across the street would be able to give is my problem, as well as the distortions and resolution will greatly effect the quality of the calculation.
Pretty soon I want to do a simple scene of a dragon landing on a famous landmark where I live. It's an asian pagoda (if that's the right term) atop a small mountain that's part of a historical site, with a road passing by it. The dragon will simply fly in overhead the camera (which will be positioned across the street looking towards the pagoda) and land on the sideroof, and will basically just act angry, roar and knock off a whole bunch of shingles (sort of like what the dragon did that Harry Potter fought in the most recent movie).
Recently I made a (very) short film where I used an HDR ball for lighting with radiosity (as the background); the position of the ball was pretty much where the CG object was, and I positioned the texture in 3D space in exactly the same spot. I also created a simple plane object that used the same HDR probe as a texture that matched the spherical projection of the background lighting, thinking that this was the correct way so the reflected light that came from the ground (which was stored in the probe image I was using) would be in the right position and have the right intensity when lighting up the shadow areas of my CG object. I offsetted the texture by the heighth of the ball in the real world, which automatically textured the plane correctly with the image probe (besides the distortions caused by the spherical mapping); sort of like camera mapping I guess. Should I really have done this? Or would simply using the background lighting have been good enough? Because when I did this it looked better.
The predicted problem that I think I'm going to face now with the dragon is that I will not be able to take photographs of the ball in the same position as the dragon will be (once it lands on the roof, way out of my reach). I was thinking of just taking them where the camera will be positioned, and doing what I did in my previous short film; but my confusion and problem is that instead of offsetting texture mapping only a couple feet (where I had thought the distortions would be negligible), it's going to be probably a hundred feet or more this time. My confusion is that when making light probes from reflective balls does the ball need to be positioned where the CG object will be? Or can you take it anywhere where it's feasible and be able to place your CG object anywhere in the HDR probe lit background and the lighting information be accurate? Shouldn't, let's say the tail end of the dragon, have a different light probe than the head end? Since they're further apart? Is there a specific distance when this technique should be used? The fact that the dragon will be sitting right on the pagoda where the bounced light from it will effect him more than what the light probe of the position across the street would be able to give is my problem, as well as the distortions and resolution will greatly effect the quality of the calculation.
