75Ard1 Lens Plus Teleconverter Experiment

Just saw your note that you resized to 1200 wide. Your Canon has a 4752 wide sensor, so if you resize to exactly 1200 your reduction ratio is 3.96. Sharpness will suffer with a non-integer reduction. Basically, none of the original image is "real", every pixel is a "figment of the algorithm's imagination". If you resize with an integer value, the algorithm averages adjacent pixels and a sharper image results. Rather than 1200 wide, I'd suggest making the final image 1188 wide. this is exactly a 4x reduction...Ray

Brandon...actually, both are showing up as 1023 pixels wide...Ray

Brandon...Usually the reduced sharpness shows up in high contrast regions or regions near the upper and lower end of the histogram. For instance, the small dark spot to the right of the S mintmark. In the 1200px image, the spot is just a spot and there is little detail in it. In the 1188 image, the spot has more detail and you can clearly see it has lighter area in its bottom right and upper left corners. Similarly, the grey dust spec at the far right of the crossbar of the first 4 just looks like a blob in the 1200 pixel image, while it has some inner tonal gradations in the 1188 pixel image.

For such a large reduction ratio, the difference is extremely subtle but I'm a sharpness hawk and this sort of thing bothers me, especially when it is easily avoided. The reason for the smaller differences with large ratio is that lots of pixels are being used for the down-sizing of each pixel in either case. The differences in sharpness are much larger for a ratio of 2 than a ratio of 4 because far fewer pixels are averaged to get the final result. And worst case is a sub-2x reduction. If a sub-2x reduction is required, then fractional reductions seem to work best...Ray

those pictures look allot better than I am getting with the 100mm canon lens and the T1i. I haven't been super impressed with my setup but hopefully once I make it to ikea in Charlotte and pick up some of those LED lights (because I can't find anything like it at Walmart around here) I will be a little happier with the pictures after that

Your last presented images are the correct width (1000px - I never noticed your note that you'd cropped to 1000 and was looking for 1200/1188) in Firefox, with no clickability. In IE9, the same images are 1060px on my screen, clickable and when clicked the result is about 1096px.

I'm using Pixel Ruler, BTW, a pretty useful tool.

When I snag your link and go directly to the Photobucket file, the image shows 1024 wide in Firefox and about 1160 in IE.

I think I'm just going to go ahead and stay confused here.

Ray, I don't know if you noticed, but it seems that the non-integer rescale also messed with the color a little bit. To my eyes, the 1200px resize is darker and maybe "more red", possibly because of averaging or softening of the image based on a non-integer resize? I did double check my process, and those images were not in any way altered color-wise. Only resized, and then cropped. Maybe this an additional interesting (possible) side effect of my bad choice of resizing factor.

I'm not having the same issues you guys are. To compare the images I clicked on each of them and then saved them locally to my PC. The saved images are both 1000x600. When I go back and forth between them the histograms and color balance don't seem to change at all.

I'm viewing the images on ViewNX, so these issues sound like they are web browser related.