Export - BMP - RGB - TFP pixel size not square.

[Allan Chesney]

I think I will have to retire so I can do all the things I would like to do!  The problem is I am away from home 12 hours a day either at work or travelling to and from. I do most of my TFP work on the train and most of my Forum replies on the train also on my phone (while something is rendering on the laptop).  That was the advantage of the FP MOV file - no other programs or tweaking involved,  just one click to start it and it rendered during the drive segment of my trip, so when I got home it was finished and ready to use.

I will try putting in a request for it and for some way to reuse the 3D radiosity 'file'. If it could be exportable maybe the old lightscape viewer would display it or maybe they can create a runtime version of the one already in TFP.  Worth a try!

Allan

[Doug.S]

Maybe if I find time I'll generate an example using TFP radiosity renders.  Pano2qtvr .mov files sizes are comparable to the .mov file sizes directly from FP.

Here I made a quick example...proves the point OK, but quality not good enough; but fixable with better "pano" image generation.

Here is what I did:
- created a one room house using basic TFP home build wizard
- added wall papers and windows/doors to distinguish each wall
- add 2 interior lights
- radiosity rendered (640x480) each wall from camera at center; rotated camera 90 degrees for each render
- combined the four images via Canon photostitch program....see attached .jpg; had to add pixels (chose gray to see the effect) to make the image format as width 2x the height  (2x horiz. x 1x vert.)...saved as .bmp (=7MB) but also made a .jpg version attached below to save file size space here.  Photostitch as you can see did a POOR job of merging the images to blend correctly to form a good panoramic image....This step seriously needs more attention....or need to render better images (camera angle views) to blend better  (I have done real photos to merge to pano and you cannot even tell they are blended; so I know it is readily possible...so my TFP camera technique needs improvement too).
- drag'n'dropped the .bmp file to the pano2qtvr "droplet" icon on desktop (see msg above in tut for link to program free source) and voila.....mov file results as shown attached below.  File size only 328K from 7MB .bmp file.  If you aim camera at ceiling or floor enough you will see image edges and the "gray" fill-in pixels....of course with a plain ceiling you could simply "fill-in" with same color as rest of ceiling and casual viewing would show OK.

So above is the method and it works....but you will need to perfect the setup.  pano2qtvr conversion took about 10 seconds to render, photostitch total took less than 5 minutes, TFP renders were about 20 seconds each at Q=4.

[Allan Chesney]

Hi Doug,

Thanks for doing the demo - as you said it shows the posibilities.  I have used the Cannon Photostich before and it did a reasonable job (from my own photos on a tripod!) but I think you cannot beat the human eye for making things look right. Manual corner joins would probably work best (might have to avoid furniture in the corners though!).

Thanks again - I will try it out when I next need that type of presentation.

Allan

[Doug.S]

Could not help but to pursue this a bit more since I need better outputs and I'm playing with making .mov files from room photos....so the techniques go hand in hand.

So I took the same one room house as above but did the following...

- set camera angle of view to 84 degrees to match the Photostich min. setting of 24mm (35mm camera equivalent) lens
- rendered at 640x640 in attempt to get more vertical coverage per rendered view so .mov would see more realistic ceilings and floors....was better but probably much better if also photo-merged renders looking up and down as well
- set camera "level" at 4', 1/2 the wall height to minimize angular convergences of vertical lines and have better photo-merge results
- needed more rendered views....used 9 instead of original 4 to improve photo-merge results....probably could use more, if more quality desired
- after each render I tried a photo-merge with the previous renders....if did not merge well (did happen a couple times), I deleted the file and moved camera and tried again...rather than hoping a merge of all at one time would work...this seems like a good approach to eliminate issues as they arise. Note I set the photo-merge program to 24mm to match the render camera FOV.
- overall I had overlap on the photo-merged image and cutoff some at each end to set a 360 degree seam at the window edge....was not perfect alignment...better trim required
- I did add to background top and bottom to get a width = 2x height image by copy paste the carpet and color fill a rectangle so in .mov mode it would look OK if viewer panned up or down very far....its OK but better would be narrower/taller renders or photo-merge top and bottom views....since TFP does not have a spherical camera.

Results below are much better and nearly acceptable....there are still some photo-merge mismatches.....somewhat a function of the images and the photo-merge tool.

[Allan Chesney]

Doug,

Yes the results are better this time and as you said more images could be used for greater quality.  It shows it can be done but I don't think it would be practical for my purposes unfortunately. Apart from the multiple image taking time etc I have a suspicion that the merged images would be almost impossible when the 'room' is filled with over 200 chairs (2,000 in one case)!  Without the chairs the whole concept would be lost.  Unless we get the MOV file back again I am going to have to make do with the AVI movie.  I suppose it can be stopped started etc to give some felling of interaction.

There are public veiwing situations where I can render the model beforehard and keep the rendered model open to "walk around" that with the audience but unfortunately I cannot save this for CD distribution.  Maybe sometime in the future it will be possible.

Allan

[Doug.S]

seems to me that an .avi viewer/player may be able to be somewhat more use-able by "scrubbing" back and forth over time as well as alternately put it into a "loop" to play continuously....but be sure to make the first and last frames exactly the same so there is no "jump" at an end-start transition.

I think some .avi players won't allow scrubbing, just playback or stop/go??  But even a "stop" and look here is better than several stills as it give a viewer some control.

[Allan Chesney]

That is the way I have been thinking on this one. Have you ever tried an AVI with the lights on? I must try it sometime but I would imagine it would take forever, especially at a decent resolution and frame rate!

Allan

[Doug.S]

I have not,
but I have done movie renders that have taken more than 24 hours to complete the render....generates a huge file (maybe 130 MB) but can be compressed to a much much smaller file (like 1 to 3 MB) via a conversion codec which re-renders/compresses in only a few minutes.

Best to always render a high quality file and save uncompressed and then use that file to compress using other codec(s) because you can't always predetermine what codec will render a really good quality compressed file and its easy to make several trial compression setups.

I just setup the render and let it run overnight with nothing else running.

Carrara comes with several codecs....just start Carrara, open the master rendered file in the render room, then save-as (re-named) using one of the codecs and you can then have a small .avi or .mov (video format, not QTVR) file for use with Macs or PCs.