About the work - December '05

The images themselves

8-bit not-calibrated .jpeg rover data

Download : MMB -- Midnight Mars Browser by Michael Howard
Processing : MMB or via PS-RGB batch-processing after Anti-vignetting processing tool, by Michael Theusner.

12-bit radiance calibrated .img rover data

Download : ftp://pds-geosciences.wustl.edu/
Processing : Daniel Crotty was very kind to help me batch-generate radiance calibrated color images for making panoramas. The final result of any mosaic I do is a contrast-enhanced view of Mars. There is an artistic input although the scene is kept as realistic as possible. False-color images are very useful in making geological interpretations, approximate true color work is as it says, approximate, though the most reliable work on true color is the work of Cornell University and Daniel Crotty.

You will notice sometimes a sky is synthetic. When frames differ greatly in brightness or when a larger portion of sky is desired a gradient fill based on two brightness levels of the actual sky is reproduced.

The stitching process -current workflow - 10-step basic tutorial

I obtained PTgui and started experimenting with the software. I took some time to get the horizon straight and realistic but once you get it, it's very easy.
Before I stitch a pan I take a close look at the frames and mosaic them in the window explorer view, using the thumbnails. For large mosaics I copy and number the frames from 001 to xxx and start stitching with the top(horizon)-row. Roughly decide what you d'like to be the center of your pan and then counting ~13 frames left of it for the first frame, frame #001..

360 ° In Ptgui;

1 settings fov = 16° full frame / panorama 360° x 90° / de-activate 'use fast transform' / projection = cylindrical or for 4-and more row panoramas equirectangular.

2 import frame 001-027 (horizon)

3 run Autopano plugin

4 Optimize -Enable fov optimalisation but check to see it's around 16°. If the result is not 'very good' or 'too good to be true' then try selecting other images as anchor image.

5 when results are still not good check if every frame has tiepoints to another frame. It frequently happens Autopano can't detect any control points when there is a big difference in brightness and shadow areas of the neighboring frames. Stitching mosaics for Opportunity involves a lot of manual selecting tiepoints because of the smoothness of the terrain.

6 If you finally get a good results-sometimes you get it immediately, you should see a curved horizon consisting of 27 frames. Using the tools in the panorama editor you can easily rotate the panorama and set the center point as close to the horizon as possible. A bit of time and practice and a straight horizon takes 1-2 minutes. There are exceptions to an easy workflow -some 360° panoramas like the 'Lookout' panorama (Spirit, sol 410-413) requires moving the center point -rotating, recentering, etc

7 Time to save the work and a copy of the PTgui project file for the horizonrow only.

8 Now import the second row's center 8 frames, run autopano

9 Optimize only the new images -use control points of the 8 new frames and those of top-row neighboring frames.
I usually disable fov optimalisation now but there are more options via individual lens parameters.

10 And so on basically -loading 8 frames at a time, running autopano, using the right images for control points and optimizing.

Some PTgui's panorama editor screenshots

Toprow of the sol 583-586 Spirit Summit panorama

16 frames on the second row

nearing completion