A Moment In Time

[Grant Atkinson]

Canon 300 f2.8, Canon 1Dmk4, handheld. 1/2500s at f5.6, iso 500. Heavy crop, as I did not want to interfere with the hunt and stayed at a good distance. The warthog mom was defending her piglets here, and they got away. So did she. My first post of this image is one of those that went missing. Due to the heavy crop, image quality is not that great.
Roger Clark of BPN kindly offerred to resize my original crop, and his resized image is a big improvement in image quality. Thanks very much Roger. I have uploaded both images for comparison. Rogers resized and improved version is my new OP above. The best I could do from a regular crop is below in the thread.
Cheers
Grant

[Grant Atkinson]

Here is the original crop.107356

[Ken Watkins]

Wherever the OP went it is a pity, I think I said that this was a good example of a real wildlife moment, in particular the BG and "incorrectly" placed bushes and branches.

Whatever Roger did to it is remarkable, I think a detailed explanation of the process would be very well received!

[Morkel Erasmus]

Roger's repost really is impressive. I know he used the Richardson-Lucy Deconvolution sharpening method, to which he frequently alludes and for which he has posted many informative links already. Roger - do you mind sharing some of those informative/descriptive threads here??

[Grant Atkinson]

Thanks Ken and Morkel for commenting. Now we can all sell those big heavy lens and just get bigger computers! Only kidding :S3:

[Rachel Hollander]

Grant - As I said in the lost thread, an incredible sighting and I love the comp. Roger's is an improvement in IQ. I really like the tension of the moment with the leopard's body position behind the bush, the warthog's tail straight up and the mud flying.

TFS,
Rachel

[Grant Atkinson]

Thanks Rachel, I won't forget the sighting in a hurry either!
cheers
Grant

[Roger Clark]

Hello all,

To clarify, resize is not really a correct term. Grant sent me a 16-bit tiff on my request and I applied Richardson-Lucy image deconvolution. Richardson-Lucy uses a model for the blur and and iteratively moves data from one pixel to adjacent pixels in order to reconstruct a higher resolution image. It is true sharpening unlike USM which is edge enhancement. USM causes bright things, like the whiskers to grow in size, whereas Richardson-Lucy makes them smaller. For Grant's image I tried several blur models and various iterations. The one that worked best was a 5x5 pixel blur and 80 iterations, which took less than about a minute on my 3 GHz I7 computer. The Richardson-Lucy algorithm I used was in ImagesPlus http://www.mlunsold.com/ (I have no relation to that company). The full resolution images shows better improvement than the smaller images posted here.

Richardson-Lucy can work wonders on some images, but it increases noise and can produce artifacts much like USM when pushed too far. Often different portions of an image need different parameters, e.g. blur is different for the out of focus areas as one moves out of the depth of field.

For Grant's image, I did some selective blurring of the background where Richardson-Lucy enhanced noise. I was impressed in the warthogs tail the original did not resolve any hair at the tip, but the deconvolved image nicely resolves hair in the tip of the tail. The leopard's whiskers came out nicely too.

A while ago Ram Mallya posted an owl image with some significant motion blur. I tried Richardson-Lucy on that image, and while it made some improvement I could not repair the blur. Grant's image here was about ideal for Richardson-Lucy (relatively small blur and high signal-to-noise image) so the result is really quite nice.

Roger

[Tom Graham]

Roger had a repost up? Not here now, can you please repost it Roger? (FWIW adding this post had quirky editting , BPN STILL having problems????) Tom

[Rachel Hollander]

Tom - Roger reworked the image for Grant and his rework is now the OP in this thread. Grant's OP from the lost thread is post #2 so you can see the difference. Other than the missing threads, I have not experienced any problems with BPN today.

[Pieter de Waal]

Hi Grant, fantastic sighting and the tension wonderfully captured. Like the environment in this one. Rogers rework of the image is equally impressive.TFS.

[Roger Clark]

Here is a before + after comparison of the leopard. The after includes a little curves to increase contrast and saturation. These are full resolution crops.

[Tom Graham]

Wow Roger, that is impressive!!!
I tried this- saved both GA original and yours (RC) then put them into PS and zoomed in some and cropped to a similar headshot size as yours. Made a screen copy in PS and put side by side. Here -
107387

I can't see so much difference - but know it is there. Not as apparent because - BPN limitation on file pixel and byte size????
Or is my comparison not valid??? Would very much appreciate your comments because so much of the comments/critiques here BPN are regarding sharpness.
Thanks - Tom

[Roger Clark]

Tom,
The reduced size BPN images have lost significant detail. Did you see my post shortly before yours from this evening with is at full resolution?
Roger

[Tom Graham]

Roger - you mean your post of 4:30PM, right? Your improvement in sharpness is very obvious, of the two leopard heads. It got me interested in further comparison but using what results from BPN requirements. What we typically see here. That is, how easy is it to see critical differences and are there really any differences. I have been accused (not necessarily falsely :S3: ) of missing/not seeing the obvious :S3: .
Many thanks - Tom
ps - I am NOT suggesting that BPN permit 3000 pixel wide images. The 1024 size I find perfect for an internet forum.

[Dumay de Boulle]

Great interaction and a very nice sighting. For interest sake you cropped from 16mp to?

[Harshad Barve]

Grant , How I wish this was in my files , terrific capture
TFS

[Mark Wiseman]

Hi Grant,
I cannot believe the difference in Roger's version. A wonderful image and a true lesson by Roger.
Thanks for sharing and best wishes for 2012.
Mark.