CompuServe Thread

#rendering regions

8 messages in this thread
#132048From: Greg MillbankOct 29, 1994 2:28 PM
I am having problems rendering regions. I am using 3DStudio version 3, and am rendering at 1500 X 1050 to a null display to a compressed TIF file. When I select a region to be rendered, my output isn't what I selected. Sometimes the rendered image is a portion of the region I selected, but not the whole region. How can I solve this problem?
#132055From: MARTIN G FOSTEROct 29, 1994 4:26 PM
Greg, >> Sometimes the rendered image is a portion of the region I selected, but not the whole region. How can I solve this problem? << I believe that's the way render region works. The region you select will only be a portion of the 1500×1050 you chose. Therefore to avoid this use a camera to completely frame the area you want and use render view instead. (Turn on safe frame to make sure you are framing correctly).
#132389From: Greg MillbankOct 31, 1994 11:52 AM
Perhaps I wasn't clear on my problems with rendering regions. My initial intent was to make a complete rendering that was a 1500 x 1050 TIF. But the model was far to large and complex for my computer to render the entire view in a decent time frame. So my idea was to keep the putput resolution the same (1500 x 1050), but this time render 4 regions of the complete intended image, ie: with safe frame turned on, I quartered the visible frame and rendered each quarter by using render region. This would leave me with four 1500 x 1050 images to be pasted together in another program. Each of the four 1500 x 1050 TIF's would look like 1/4 image and the other 3/4 black. This should have worked. Instead each of the four TIF's was not the complete REGION I selected. Each was just a PORTION of the REGION I selected. Is there some setting I have to change?
#132663From: Forcade & AssociatesNov 1, 1994 12:52 PM
>> Perhaps I wasn't clear on my problems with rendering regions. My initial intent was to make a complete rendering that was a 1500 x 1050 TIF. But the model was far to large and complex for my computer to render the entire view in a decent time frame. So my idea was to keep the putput resolution the same (1500 x 1050), but this time render 4 regions of the complete intended image, ie: with safe frame turned on, I quartered the visible frame and rendered each quarter by using render region. This would leave me with four 1500 x 1050 images to be pasted together in another program. Each of the four 1500 x 1050 TIF's would look like 1/4 image and the other 3/4 black. This should have worked. Instead each of the four TIF's was not the complete REGION I selected. Each was just a PORTION of the REGION I selected. Is there some setting I have to change? << You would have to set your output resolution to 3000×1050 and then do region to make this work, which wouldn't save you very much. You could use the technique and do a render/blowup, but it will be hard to get the same size area selected each time. If you do not, you will get different scales for each section. One other method is to use a camera for each of the 4 sections. You will need to create a camera that covers slightly more than a quarter of the image area. then shift-move that camera to each of the other three sections. You would then have overlapping camera for each section and you can paste them together in Photoshop or the equivalent. It might help if you built-in a border with markings that will help you to align the parts. This is going to be a lot of work and has lots of room for error. If 3ds had a way to specify the blowup region by numbers it might work. Your best bet is probably to go ahead and do a straight render to the full size. It will probably not really make that much difference in render time and will free you up to do something productive while you're waiting.
#132740From: Greg MillbankNov 1, 1994 7:28 PM
Thanks for your advice, but to tell you the truth, it makes a huge difference in Rendering time. We are currently using a 486/66 with 128 MB of RAM. Rendering the whole model in one pass causes the computer to begin to swap to disk. This means 1 to 1.5 weeks of rendering time. Done in parts, it takes about an hour and a half or two per section, for a total of less than a day. (Right now, more like two days because of the screw ups we are trying to avoid)
#132939From: Forcade & AssociatesNov 2, 1994 5:49 PM
>> Thanks for your advice, but to tell you the truth, it makes a huge difference in Rendering time. We are currently using a 486/66 with 128 MB of RAM. Rendering the whole model in one pass causes the computer to begin to swap to disk. This means 1 to 1.5 weeks of rendering time. Done in parts, it takes about an hour and a half or two per section, for a total of less than a day. (Right now, more like two days because of the screw ups we are trying to avoid) Thanks again. << 1.5 weeks vs 2 hours! That's quite a differential. Have you looked at using similar overlapping cameras to get your sections? Terry Gilbert Forcade & Associates
#133118From: Forcade & AssociatesNov 3, 1994 10:35 AM
>> Thanks for your advice, but to tell you the truth, it makes a huge difference in Rendering time. We are currently using a 486/66 with 128 MB of RAM. Rendering the whole model in one pass causes the computer to begin to swap to disk. This means 1 to 1.5 weeks of rendering time. Done in parts, it takes about an hour and a half or two per section, for a total of less than a day. (Right now, more like two days because of the screw ups we are trying to avoid) Thanks again. << Is there any way to resample some of your texture maps down to make up for the increased image size? Could you render selected objects and then composite them together? Are your maps using Summed Area or Pyramidal filtering? Are you using two texture maps for any materials and could you reduce that to one composite texture? Just some questions that might have an impact on memory use. Terry Gilbert Forcade & Associates
#132120From: EDWARD HARVEYOct 30, 1994 1:02 AM
Try render blowup. It will blow up any area you select into the resolution you specify.