Closed

Large Photoshop image retouch based on series of Gigapan images

This project received 11 bids from talented freelancers with an average bid price of $558 USD.

Get free quotes for a project like this
Employer working
Skills Required
Project Budget
N/A
Total Bids
11
Project Description

The images in the source folders (to be sent on acceptance of the job as they are large) are all photos that will be stitched together to produce a 3D panoramic scene (approx 300 tiled shots in each panorama) Sample images of the scene have been uploaded into this job description.. However while shooting, the shutter speed differed across most photos, such that they ranged anywhere from 1/8 of a second to 1/520 of a second. The shutter speed severely affected the amount of light that entered each photo, such that when stitched together, the panoramas were differently exposed blocks.

To solve this problem, we started normalising each of the photos in Adobe Lightroom 4 (Lightroom was chosen for it's non-destructive editing). We reviewed the shutter speed of each photo, and calculated an exposure modification based on that. We used Lightroom's exposure change slider and edited each photo, such that the exposure was 4 times the difference between 1/40 and the image's actual shutter speed.

So, if an image's shutter speed was 1/25 of a second, it differs to the nominal 1/40 by -15 (40 - 25). We multiplied that value by 4 and set the exposure adjustment to that, so -0.6. Some examples:

1/8 = [url removed, login to view]
1/40 = No change
1/80 = [url removed, login to view]

Note that values over +[url removed, login to view] blew the images out too much, so no images were set higher than 2.00. There were also points where the regular calculations were modified as they were actually not correct, so the above tends to be a common correct value, but not always.

After running through the range once, we stitched them together using Gigapan's Stitch tool ([url removed, login to view]), and saw that there was a massive improvement, but now we need to improve it further. This must be done by further adjusting the exposure levels in Lightroom, and then by making adjustments by hand in Photoshop. Hand adjustments include finding images of high contrast (usually an image which contains a chair / couch and a white background (tiles or curtains)) and manually uncontrasting them.

So, what we need from here is the following (for each scene):

1. Use Adobe Lightroom to further tweak the exposure levels in a non-destructive way (using the exposure slider in the Develop section)
2. Making adjustments by hand to correct areas of high contrast, and to sharpen some images that ended up being blurry
3. Exporting the files using Lightroom's Export to Hard Drive feature (this keeps the original files safe, with the non-destructive edits in place and creates a second copy of the files to use in the stitch).
4. Testing the exposure levels by importing the scenes in Gigapan Stitch and stitching the files together, enduring that there is no noticeable blockiness to the exposure levels.
5. Repeat from 1 if the results are unacceptable.

And after that:

1. Export the stitched Gigapan panorama to .tiff, and edit the .tiff in Photoshop
2. Note that Scene 1 was 26 columns, Scene 2 was 30 columns and Freezer is 5 columns (Gigapan needs to know this to lay out the scenes prior to stitching).
3. Scan over the entire .tiff, expertly and cleanly editing out blurry parts, fix where the image wasn't stitched together properly, and generally clean it up so that we can zoom almost all the way in and see a great result (note that shading, blur and even the noise must be pretty consistent, such that if a touch-up affects these, we'll need to match it back into the document).
4. Remove the 'X's from the curry box on the table
5. Remove the 'X's from the food boxes in the freezer
6. Save the .tiff, and send it, plus the Lightroom Catalogs and images back to us.

Looking to make some money?

  • Set your budget and the timeframe
  • Outline your proposal
  • Get paid for your work

Hire Freelancers who also bid on this project

    • Forbes
    • The New York Times
    • Time
    • Wall Street Journal
    • Times Online