Garden Gnome Software

I find myself doing a lot of back forth converting. I'll start from equi and convert to cube - do a few edits and covert back to equi. I keep everything in tif format but notice stitching blurs in the places where the cube conversion cuts the image.
Is this normal or is it better to covert differently. PTgui creates equi - cube conversion is handy for editing but equi is more user friendly output for using in 360cites or other media that favor the file type.

thanks
Bill

Welcome to the forum Bill [Nodal Ninja],

We find that most stitching errors come from not having your NNP calibrated properly. But that is probably not your issue if you are using a Nodal Ninja panorama head. IF there are any major stitch issues... we go back to Ptgui and try again for a clean stitch. Any issues after that are cleaned up in Photoshop before bringing our 6000 x 3000 16bit Tif into Pano2vr. We do convert our equirectangular image 1 time into separate cube faces for final examination and tiny stitch issues [if there are any remaining]. At this work flow point we also place our clients logo into the nadir.

When we get ready to convert our cube faces back into a equirectangular image... the setting shows the final image to be 5999 pixels wide. So we change this value to read 6000.

Its an internal rounding error:

Equi to cube face 6000 /Pi (3.1415) = 1909.91564539
Cube face to equi 1909.91564539 x 3.1415 = 5999.9999999

So we change this value to read the original stitched image 6000 pixel value.

Maybe when you do repeat conversions without adjusting to 6000 the process is creating your issues.

You can keep the inevitable (!) re-mapping damage to a minimum by doing the following:
1. make your initial equirect in PTGui.
2. convert it to cubefaces, but also keep the original equirect.
3. do your editing on the cubefaces.
4. reconvert them back to a new equirect, making sure its dimensions are identical to the original one.
5. load the original equirect in Photoshop (or Gimp), place the reconverted new equirect in a layer on top of it.
6. temporarily change the layer blending-mode to 'difference', so you can see where the repairs can be found.
7. make the layer containing the new equirect invisible by completely masking it, then erase the mask locally to reveal only the repairs.
8. merge both layers, and save.

Maybe this reads like a lot of work, but after a bit of practice you'll see it just takes a few minutes.

Erik

nodal ninja wrote:I find myself doing a lot of back forth converting. I'll start from equi and convert to cube - do a few edits and covert back to equi. I keep everything in tif format but notice stitching blurs in the places where the cube conversion cuts the image.
Is this normal or is it better to covert differently. PTgui creates equi - cube conversion is handy for editing but equi is more user friendly output for using in 360cites or other media that favor the file type.

thanks
Bill

First thing to do is only do the conversion equi -> cube -> equi once.
Another way to make edits in the image at poles is to use the numeric transform in PTGui to position the nadir in the middle of the horizon. Output the transformed equi -- do your edits then reload the transformed equi back into PTGui using equirectangular as the lens type (first tab) then use the numeric transform again do put the nadir (or zenit) back at the pole.

So you never work directly with cube faces. You can output transformed layers and use those also if you need to.

Another way to make edits in the image at poles is to use the numeric transform in PTGui to position the nadir in the middle of the horizon.

For those who'd like to know: this is done by manually 'pitching' the equirect + or - 90 degrees in the 'Image Parameters' tab.
A disadvantage of this method is that there will hardly be a straight line in the area to be repaired, making things a bit difficult in some cases.

Loss of quality for the entire equirect after pitching it up and down (for the nadir) + down and up (for the zenith) is comparable to the cube face method though, a remap is still a remap.
So in order to keep visible damage to a minimum you should go through the same procedure in Photoshop as for the cube face method I described.

Erik

gee - from internal rounding errors to numeric transforms to manually pitching the equi - this is too easy
Seriously though I genuinely appreciate the time you guys spent in laying out these suggested work flows and tweaks. It makes forums like this valuable to those of us with lesser skills sets in the software department. So now I need to roll my selves up, kick the dogs out (gently or I might loose a foot), close the office doors and get stuck into developing a workflow that works - I'll start with the easiest and work up from there.
Glad to be part of the forum -
Many thanks!
Bill

nodal ninja wrote:I find myself doing a lot of back forth converting. I'll start from equi and convert to cube - do a few edits and covert back to equi.

Hi, Bill!
Why do you do convert to cube? I also did it before to fix the nadir, but now I use the patch tool in Pano2VR. With it you can reduce the part of the image to the necessary and then "insert" it again in the original equi. You can even convert it again into a new equi with the patch in it. Obviously that tool does not work only for the nadir, I sometimes choose a small part of the equi to fix it with it.

Regards, Martin

BTW. Congrats for your new tools at NN. I hope I can get one of those soon.

Hey Martin,
I like to convert to cube when needing to do edits on the outer edges of the equi - like for instance the nadir. Center stuff isn't has important because the distortion is minimized. The patch tool is great but more often that not I need a tad bit more editing. Refining my workflow in the field as well will certainly reduce workload during post-production.
thx
Bill

ps - thx on the congrats - lots on the horizon

nodal ninja wrote:gee - from internal rounding errors to numeric transforms to manually pitching the equi - this is too easy
Bill

Yeah, sounds gnarly -- but do any of the methods once or twice and it gets real easy.

I use supercubic to fix the poles in photoshop and try not to convert to cube faces at all apart from that.

http://www.superrune.com/technical/soft ... rcubic.php

Yep i have to agree with ambientlight on this.
Superrune pluging for photoshop is a great little tool.
best not to convert back and forth if it is not needed.

Alex