(e-mailed) alexchen - 95/100 + 1 bell (artifact winner) readme - no bells and whistles artifact a little blurry/ghosty - 1 point AccumulateBlend assumes right to left order works only when at most 2 images overlap in any one pixel - 1 point alpha channel has values > 1 doesn't take into account existing alpha (curved edges) - 1 point BlendImages max_x was never computed (not used, but still...) - 1 point warp x shift only works for left stitching y shift ok y warp assumes left stitching and y_min is last image - 1 point (e-mailed) animura - 93/100 no readme artifact has pretty big double-images - 2 point l-k output NG uses original img at each level, not pyramid level p[lvl] - 5 points stitching ok (e-mailed) camden - 96/100, 1 late day no readme artifact late: 2/10 - 1 late day has a little bit of blurring lk output just a tiny bit different Ix, Iy, It computed backwards! but it all cancels out - 3 points matrix solved by substitution, a little less precise - 0 points (didn't use Kramer's rule) stitching has wrong vertical bounds final warp matrix is not quite right - 1 point (e-mailed) cvana - 94/100 readme - attempted to auto fix brightness (whistle) + 0 whistle artifact has brightness problems, big blurs - 1 point did not answer questions - 3 points lk output off by small but significant amount used sobel operator for Ix, Iy OK stitching - bbox looks off right-stitching has serious skew and blend problems changed function signature of AccumulateBlend OK AccumulateBlend assumes right to left, maybe others -1 point affine deformation includes scaling, why? - 1 points (e-mailed) dane - 99/100, 1 whistle readme - panorama includes moving person in multiple frames + 1 whistle artifact good! blending has skew and blend problems strange blend approach... I think will fail right to left - 1 point affine deformation assumes right to left (e-mailed) dbb - 94/100 no readme artifacts ok, small amount of blurring at far left - 1 point lk off by large amount sign errors in computation of Ix, Iy - 5 points blending ok but bbox, vdrift not right vdrift compensation will not work for right-stitching - 0 points (computed leftmost and rightmost indices but never used them!) (e-mailed) ehwang - 98/100, 1 late day, 2 whistles readme - 2 whistles - same person multiple times, + 1 whistle modified with scissors and photoshop + 1 whistle artifact late - 2/10 - 1 late day artifact has major blend/offset problems - 2 points turned in a zip instead of individual files (and no executable) - O points lk off a bit used sobel affine correction wrong for right stitching - 0 points alpha empty (e-mailed) gerek - 98/100, 1 late day readme - no B&W artifact late - 2/10 - 1 late day code late - 2/10 lk off tiny bit dt[1] computed wrong, and inverted - 1 point ATb is inverted - 0 points bbox or warp looks off (global vertical shift) y translation amount is off - 1 point (e-mailed) gkhuang - 98/100, 1 whistle readme - B&W - panorama caught same person in multiple frames + 1 whistle artifacts have some ghosting & blend artifacts - 1 point everything looks good except vdrift for right stitching assumes left stitching (but close) - 1 point (e-mailed) jderus - 98/100, 2 late days + 1 bell (artifact winner) readme - no bells or whistles some artifacts 2/11, others 2/12,2/13 - 2 late days blending & ghosting artifacts - 1 point code 2/11 vdrift is not quite right (guess and check) - 1 point (e-mailed) jglouie - 94/100 artifact had some big blending/ghosting problems, but fixed when stitching done in the right order - 0 points lk off by large horizontal amount cramer's rule applied incorrectly for x: x -> (bq - pd) / (ad - bc) - 5 points vdrift assumes left-stitching - should be over cShape, not mShape - 1 point crop isn't quite right he did notice that stitching done in wrong order no bells and whistles (e-mailed) johnnyho - 97/100, 2 late days readme - no bells and whistles artifact 2/11 - 2 late days big blend/ghosting artifacts - 2 points code 2/11 y translation computed wrong - 1 point (e-mailed) jtyrrill - 96/100, 3 late days readme - used all 3 late days - 3 late days artifact 2/12 (update 2/13) looks great, but didn't answer questions - 3 points code 2/12 vdrift wrong on right stitch uses mShape bounds instead of cShape bounds - 1 point vshift assumes left stitch - 0 points (e-mailed) kierah - 97/100, 2 late days + 1 bell (artifact winner) readme - no bells or whistles artifact 2/11 - a little after midnight - 2 late days a little ghosting - 1 point code 2/11 code turned in was skeleton (release) but debug was ok - 0 points start_x and start_y are wrong (don't use minx, maxx) - 1 point vdrift wrong - only works for pairs of images (looks at n-2 and n-1) matrix looks like guess and check crashed on right-stitching - 1 point