Forgot your password?
Dave Donley, September 6, 2003 in Rendering
I'll try to.
Here's how I would (try to) do it:
1. Grab one of these cross pano images:
2. Split it up using screenshots into six perfectly square image files.
3. Create six image props and situate them so they are way big and surrounding the origin, in a cube. They should not be plugins, and they should use the Constant reflectivity shader.
4. Create an array of say 10 x 10 point lights in a square array on each face of the cube.
5. Set the color of each light to that of a pixel within its portion of the image; it might be useful to re-export the image out in a 20 x 20 pixel size to average out the colors automatically. I would use the magnifying glass of the Mac OS X color picker to get the image pixel colors to use for the light colors.
6. For the very bright areas the light brightness would somehow be set higher than 100%. The light brightness would be where the high range value would go. To be simple I would just set the brightest areas to some high value. This part would be done in a more sophistuhmicated way to properly do HDRI files.
7. Convert each point light into a directional light. The direction vector would be set as a the vector from the point light location to the origin, so the cube were centered on the origin, the light directions would be set correctly when converted.
7a. Select all the lights and make a symbol out of them.
8. Render the big image prop cube without the model in it. Set the view from somewhere in the cube. Fiddle with the view until the distortion is minimal (may have to move the camera vertically to get the height to match that of the source panorama). Take a screenshot that exactly matches the page bound rectangle. Create an image background from the screenshot.
9. Create a new layer and try to approximate the same scale and view that is shown in the image background. It should be enough to save a sheet that saves the view but not the layer. That way you can restore the view on your new layer to that of the previous layer.
10. Put some cool model in the model layer. Some amount of mirror reflections would be nice. (Actually you won't see proper reflections of the images. To do this right we would have to place the props in the model layer and somehow set them to not cast shadows.)
11. Set the model layer's background to the image background. Place the environment light symbol in the layer. Turn off the layer ambient light. Render. It would be possible to use mapped shadows with this rendering because we are not losing shadow map precision because there isn't a huge ground plane, just the model.
[ 09-05-2003, 01:49 PM: Message edited by: Dave Donley ]
MikeB, can you post some results ? Maybe also from the wireframe model so I could understand how it's supposed to look like. I'm anxiously waiting! (sorry for my english)
Sorry, It'll be awhile befor I can try this. I'm in a huge production push both in my full time job and in my side work. I was just curious about the technique. Maybe Dave can post some screen shots.
I tried taking some panoramic pictures near my house that I am going to use. I was going to get a silver Christmas ornament from the basement and try that as well, but we only had green, red, and a milky white color of ornaments.
With so many lights this would really need to be automated. I would like to write a plugin that asks for a cross-shaped panoramic image and will create arrays of lights from that at whatever resolution you would want. That would be a big start.
I am trying - but failing - to understand this:
What is HDRI - in fairly simple language?
HDRI = High Dynamic Range Image.
Normal bitmap have 8 bits per color channel. An RGB has a total of 24 bits, or 32 bits if you would include an alpha channel.
HDRI can contain more information than 8 bits per channel. First of all, it can exceed the 256 samples of a classical color channel. Second, it's also possible to use floating point.
The advantage is that HDRI images can contain more info, which is not directly visible AT FIRST SIGHT. But it would be visible if, for instance, you would change the brightness. A good link would be:
The advantage in rendering is that reflections still respect the "natural" intensity and contrast of light sources, which leads to more natural speculars. Another advantage is that you can easily use it as a "light map": wrap it around your scene, use the intensity and color to project light into your scene, and presto: you have a very "real" scene.
It's not possible to "Fake" HDRI in VectorWorks: you would be missing the High Dymanic part of the deal.
Thank you. I understood that :-)
You can post now and register later.
If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.
Pasted as rich text. Restore formatting
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.