Resampling quality difference between IMatch 3.6 and 5.1.6?

[jonz]

If I turn off all sharpening and downsample a large image in the batch processor to a 600px wide TIF I am seeing an appreciable visual softening in the IMatch 5 version when compared to the same image downsampled in 3.6. This is using bicubic sampling (no sharpening) in 3.6 and resizing to the same dimensions in v5 with no sharpening enabled. It's enough of a problem that I need to go back to version 3.6 because the photos just look mushy.

Has anyone else seen this? I believe I'm keeping everything constant except that in v5 I don't have a choice of bicubic ... it's one flavor resample only. I am not using the image cache image.

[jonz]

I am uploading two samples, both tifs, to show the difference.

[attachment deleted by admin]

[jonz]

Am I the only one seeing this? To me it seems like the batch processor works well resizing color images, but the softness comes in when downsampling black and white.  Since a lot of my images are b/w that's a problem for me, and I haven't been able to find a solution except either going back to 3.6 or outputting a full-rez version and resizing in LRoom, which I'd like to avoid.

[Mario]

If I turn off all sharpening and downsample a large image in the batch processor t

The batch processor uses the same rendering queue as the rest of IMatch. IMatch 3 always added some sharpen. When you resize an image and you don't sharpen it a bit, it will look soft. This is how the resize algorithms work. The result depends on the ratio between the original image and the resized version, and the resampling algorithm you choose. Why do you turn off all sharpening?

[jonz]

OK, this starts to explain since I thought the straight "bicubic" choice in 3.6 involved no sharpening.

I will work on unsharp masking (in 5) and see if I can get the b/w output to where I like it. Am I correct in understanding you to say that all downsampling in Version 5 is bicubic no sharpening, and that all sharpening is controlled by the USM settings? That would make sense...

In answer to your question about why I had turned off USM (in 5) because I was trying to do an apples-to-apples comparison between 3.6 and 5. But what you are telling me is that even with only bicubic selected in 3.6, there was still sharpening applied (which I never guessed because the other settings specified sharpening). No wonder the results are different!

[jonz]

Working with the USM settings has gotten me pretty close to where I want to be. Thanks for the help.