Terragen truncating camera keys on save, causing camera judder?

[Hezza]

Hi there,

We've been using terragen in conjuction with maya the past couple of months. The method so far has been to export a ter file from terragen, import that into maya, setup cameras and then export the cam back into terragen as .chan files. Apart from some slight offset problems this process has been managable but something strange has popped up.

Basically we've exported a camera as a chan file, imported it into terragen and everything is fine, the camera moves smoothly on playback, but once the file is saved and loaded again the camera movement becomes very juddery.

-------------
chan file import
-------------
1   16360.58906   1573.920086   -12548.31302   -6.599767171   -23.27797665   0   64.47
2   16360.45419   1573.910312   -12548.2362   -6.616518573   -23.32380417   0.02085491917   64.47
3   16360.31964   1573.900537   -12548.15954   -6.633899263   -23.37135328   0.04249328152   64.47


----------------
terragen Values after Save
----------------
16360.6   1573.92   12548.3   -6.59977   23.278   0
16360.5   1573.91   12548.2   -6.61652   23.3238   -0.0208549
16360.3   1573.9   12548.2   -6.6339   23.3714   -0.0424933

Any clues on how to combat this?

Cheers
Hez

[cyphyr]

It may not be relevant but I have noticed a similar "feature" when drag 'n' dropping colour swatches in Terragen.
You can drag a colour swatch on to another colour swatch (in a power fractal for example) but in doing so the value becomes "truncated".

For example (in power fractal):
a value of blue (R80, G145, B1970) 0.4151 (on the slider)
when dragged goes to
a value of blue (R80, G144, B168) 0.3993 (on the slider)

I think Terragen truncates value but I dont know by how much.

Richard

[Matt]

The problem is with the value being rounded down to 6 significant figures when saved to the TGD file. If the values were much closer to 0 this wouldn't really be such a problem because the 6 significant figures would represent smaller numbers. The old fashioned way to get the most accuracy in this situation was to keep camera motion close to the origin and translate it to the final position somehow. Unfortunately there's no way to do that in TG2 unless you could translate the whole scene, and I doubt that's a practical solution.

If you are rendering from the user interface, the obvious workaround would be to press the import button on the camera before rendering, but if you're rendering remotely that's not possible.

I should be able to increase the accuracy of the numbers saved into the TGD. If the email address in your profile is correct I'll email you with something tonight.

Matt