CS766 Project 1 Report

Shijin Kong (krobin)

GOAL

This project aims to recover high dynamic range radiance maps from a series of photographs taken with different exposures. Debevec's algorithm is used to calculate response curve and further recover radiance map with it.

INSTRUCTIONS TO RUN PROGRAM

• I wrote a command line executable program which implements the following functions:

1. Use gil library to read in a set of photos and read in configuration file, including the exposure for each photo.
2. Pick certain number of pixels from a set of photos. Those pixels are ensured to have wide range of pixel values on the photo set as to make recovered response curve more accurate.
3. Use CLAPACK library to perform Singular Value Decomposition (SVD) on a group of over-determined equations, find the least-square solution and then get the recovered response curve.
4. Output the high dynamic range radiance maps (.hdr file) with recovered response curve.
   

• The executable file is HDR.exe, which takes two parameters, you can run HDR.exe to see help:
  
      >HDR.exe
      Usage:
      HDR.exe img_path img_prefix
  
  The first parameter img_path indicates the directory that photos are stored, the second parameter img_prefix is the common prefix for all photos in that directory. After running HDR.exe, there will be a img_prefix.hdr created under img_path. For example, if running HDR.exe on photos inside photo1 directory (including photo1_01.jpg, photo1_02.jpg, ..., photo1_13.jpg), please input:

          >HDR.exe photo1 photo1_

     The output results is stored as photo1/photo1_.hdr

PHOTO SETS

I took three photo sets, which are stored in directory photo1, photo2, photo3 respectively. Besides photos, each directory contains a filename called "shutter.txt". The first line of shutter.txt is the number of photos in this photo set, indicates how many photos will be used. The next few lines are exposure time for each photo. Let's look at photo1/shutter.txt as an example.

photo1/shutter.txt:

13               // <-- total 13 photos in photo1 directory
1.00000          // exposure of photo1_01.jpg is 1.0 second
0.80000          //
0.40000
0.20000
0.10000
0.05000          // ...
0.02500
0.01250
0.00625
0.00313
0.00156
0.00080
0.00040          // exposure of photo1_01.jpg is 0.0004 second

HDR RESULTS

• A Tonemapped JPG file and .hdr file is already generated in each photo set directory. Name of tonemapped and .hdr files:
  For photo1 set: photo1_tonemapped.jpg and photo1_.hdr
  For photo2 set: photo2_tonemapped.jpg and photo2_.hdr
  For photo3 set: photo3_tonemapped.jpg and photo3_.hdr

• Show photo1 set as an example, the photos I took:

     

  The tone-mapped picture of photo1:

  

  Response curve is like this:

  

IMPLEMENTATION DETAILS

• Everytime HDR.exe is running on a photo set, it first reads "shutter.txt" file from the set directory to find out image count and their exposure times. The program then picks 50 pixels separately for each channel and solve the optimization problem.

• Algorithm on how to pick out 50 pixels for each channel. I was trying to find pixels with as wide range as possible.
  Step 1: Set initial threshold T (only a little smaller than 255, I used 230), and say we have N photos, the first image has largest exposure.
  Step 2: Scan photos. For any pixel, if the difference of pixel values of image 1 and image N is greater than T, then record that pixel.
  Step 3: After scanning, if more than 50 pixels are recorded, then we are done. Otherwise, set T = T - D (I used D=20) and go back to step 2.

FAVORITE ARTIFACT

It's the photo1_tonemapped.jpg I list above, or you can get it from here.