I also get these artifacts as someone else mentioned. I can't use 16-bit raw, photoshop messes up the image when I use that mode. Also Carrara fails to import from that mode (Import error). So instead I saved the image as a 16-bit TIFF, that works.

It looks okay in Carrara in the model, but when rendered it looks like this:



Could the reason be that I used a OTA height map (exported from Annihilator), a bmp in resolution 162 * 360, which I scaled up to 2592 * 5760 and 16-bit greyscale.

to me the grid things look like stuff you get when upscaling images. What algorythm did you use? ie in irfanview you can chose different ones. most simple one is just "doubleing" pixels which will make the large image "blocky" while others algos interpolate between pixels for a "smoother" image

I used Bicubic (smoother when I enlarged it), but then I had to reduce it once more to 4097, because it seems carrara doesnt handle bigger files, and then I used bicubic instead of bicubic sharpening (which Photoshop says is best for reduction).

But the model looks okay in Carrara, and also when rendered with the default camera (I think it's conical). Here's a sample (yeah, those are supposed to be mountains):

Lower the bump in global shader. Also, if import is still blocky, use gaussian blur in photoshop. Also, remember to use 16bit, single channel, intel byte order, no header RAW's with both photoshop AND carrara AND mapconv.

I did fix the issue somewhat by applying gaussian blur in photoshop: I had done it before too, but then with 2 pixel radius. Now I applied one with 5 (or was it 6) pixel radius and it helped.

I used a bump with the size of 0,01 ft in Carrara.

Single channel means greyscale I assume?

I used TIFF:s since I couldnt get raw working in neither Photoshop nor Carrara: when using raw, photoshop doesnt reload its own image well at all, even when supplied the same information to the raw as it was saved with. I used Photoshop CS3 by the way, is there anything better with newer versions of it?

I use cs3, works fine with raw for me.

Well, that's weird. Maybe it depends on map dimensions then. my last map, where I managed to use 16-bit tiff was a 6 x 12 in spring, that makes it 3072 x 6144 pixels large in texture map (scaled to 1844 x 4097 before import to Carrara). There the raw method didnt work at all, I got an import failed-error.

Now with another map, with spring dimensions 12 x 10 and a texture map with the dimensions 6144 x 5120 pixels gives me other results: now I don't get the import failed errors but instead the tif-map is imported as flat (also commented in this) forum). Raw and photoshop formats give me weird results, basically another map than what I made, also depending on the dimensions (if scaled to a 4097 x 3414 or 2049 x 1707 pixels before importing to carrara).

Here are some examples:

1) TIFF-format, 8-bit, dimensions 4097 x 3414. Result: as expected. When rendering gets scanlines though.




2) TIFF-format, 16-bit, dimensions 4097 x 3414. Result: flat




3) Raw-format, 16-bit, dimensions 4097 x 3414, Intel byter order, header 0. Result: scrambled




4) Photoshop format, 16-bit, dimensions 4097 x 3414. Result: Interesting (but altered heightmap)




5) Photoshop format, 16-bit, dimensions 2049 x 1707. Result: scrambled.




6) PNG-format, 16-bit, dimensions 4097 x 3414. Result: as expected (but may not be optimum for rendering)




So for this heightmap, the png-format was the only 16-bit format that worked for me. Hopefully this detectiove work is of use to someone else too (I like detective work).

Finally sorted this issue: the map needs to be quadratic in order for the raw format to work. So I had to stretch the 2049 * 4097 map to a 4097 * 4097 format. Hope I didn't throw out the baby with the water.

16-bit and 16-bit per channel are different formats btw.

I assume people mean 16 bits per channel when they talk about 16-bit raw format. But since it's a greyscale anyway it's the same in this case.

Was more for noobs who come in later and try to read up on it.