| Author |
Replies: 48 / Views: 6,416 |
|
Pillar of the Community

United States
4038 Posts |
In my quest for ultimate sharpness, I sent my Canon T2i to have the Anti-Aliasing filter removed. It's now "High Resolution" or "Hot Rodded". I figure the combination of HRT2i with the 105PN at largest aperture ought to give me the sharpest image I can get for Cents at around 0.8:1 mag, so my first test subject is the 56-D I've been imaging for a while. Hopefully Photobucket doesn't mess it up. You can supposedly view full resolution image on my PB page. Image below is an 8-image stack with the 105PN at f/2.8. With M=0.8, this brings effective aperture to f/5, well below the f/6.8 diffraction limited aperture of HRT2i. I could actually stop down the 105PN to f/3.3 and still be within the DLA limit, but the 105PN is sharpest at f/2.8 so only disadvantage is a few extra images in the stack. I may try this again at f/3.3 to see if it makes a difference, but here is the f/2.8 stack for now: 
|
|
|
|
Moderator
 United States
23522 Posts |
Quote: Prepare ship for Ludicrous Speed! You've reached the point where the only way to tell the improvement is to display at 100%.  My monitor has 2560x1440 in-hand. Can you link the fullsize image? I want to see this. 
|
|
Pillar of the Community
  United States
4038 Posts |
Contact me for photographic equipment or visit my home page at: http://macrocoins.com
Edited by rmpsrpms 04/14/2013 7:38 pm
|
|
Moderator
 United States
23522 Posts |
Done, and....words fail me. I, literally, cannot think of anything to say which describes what I'm looking at, save this one thing:
You've changed the game, Ray. Coin photography was one thing before you aimed your attention at it, and it's a completely different thing now.
|
|
Pillar of the Community
  United States
4038 Posts |
Note the stacked image is a bit over-sharpened by CZP. The program does a final sharpening that is more aggressive than I'd normally do. Here is a source image direct from the camera, focused on the high points of Lincoln's beard, compared with the final CZP image:  
Contact me for photographic equipment or visit my home page at: http://macrocoins.com
|
|
Pillar of the Community
  United States
4038 Posts |
Thanks SD, it's been an interesting path to find ultimate sharpness. The source image above actually impresses me even more than the final stack, and if I had a better stacker that didn't oversharpen I'd use it. That source image is literally with zero sharpening and it's showing all the fine detail without being cartoonish.
|
|
Moderator
 United States
23522 Posts |
I'm guessing the CZP image is the top one; that's what I see in the downloaded image. It's my opinion that the "overzealous" sharpening is actually a good thing under the circumstances. The resulting image, to my mind, is more faithful to what the coin would actually look like under equivalent visual-only magnification, with all its' dings and warts starkly revealed. It's the kind of detail you're looking for with an image of that size; with a slight lighting adjustment there's no need for any additional magnification for all but the most subtle of die varieties.
Even better, downsized to a postable image size, it softens slightly if only because the details exceed the DPI of the downsized image, and even if they didn't the pixel pitch of a monitor isn't up to the task anyway.
So you get the real truth of the coin at both ends. It's as close as it gets to what someone with extremely sharp vision and a high-quality loupe would see, in-hand.
|
|
Pillar of the Community
United States
2661 Posts |
 I am in awe. I think you may have just achieved perfection. Best I have ever seen.
|
|
Pillar of the Community
  United States
4038 Posts |
Quote: I'm guessing the CZP image is the top one; that's what I see in the downloaded image. It's my opinion that the "overzealous" sharpening is actually a good thing under the circumstances. The resulting image, to my mind, is more faithful to what the coin would actually look like under equivalent visual-only magnification, with all its' dings and warts starkly revealed. It's the kind of detail you're looking for with an image of that size; with a slight lighting adjustment there's no need for any additional magnification for all but the most subtle of die varieties.
Even better, downsized to a postable image size, it softens slightly if only because the details exceed the DPI of the downsized image, and even if they didn't the pixel pitch of a monitor isn't up to the task anyway.
So you get the real truth of the coin at both ends. It's as close as it gets to what someone with extremely sharp vision and a high-quality loupe would see, in-hand. That's pretty much what I've been after, a good looking overall photo that you can dig down into to see details for grade and variety attribution. With this result I think I've finally figured out just how good the 105PN is...I started using it with the Nikon D5000, and it was the sharpest thing I'd seen by far on that camera. When I upgraded to the D7000, I saw very little improvement but didn't know why. Then I learned about Nikon's shutter shake problem, and proved it was the culprit using delayed lighting, which is very impractical for coins. Based on your input the solution to shutter shake was going with a Canon that has EFSC, so I bought the T2i, which satisfied me for a long time. Yet knowing the AA filter was there blurring every shot needlessly caused me constant low level grief. Removing the AA filter had far less effect than going with Canon/EFSC to improve the image, but it did take away that last bit of blur and made the 105PN really shine.
Contact me for photographic equipment or visit my home page at: http://macrocoins.com
Edited by rmpsrpms 04/14/2013 8:43 pm
|
|
Pillar of the Community
  United States
4038 Posts |
Quote: I think you may have just achieved perfection. Best I have ever seen. Thanks Tim! I don't know how to do better. One thing I forgot to say is that this was taken with a new lighting arrangement. The 'Cap used my "Smile Director" concept to create his own type of diffuser that sits on the stage with the coin and shines light on the coin similar to the Smile Directors, but without having to modify the lights themselves. Well, one of my Jansjos went flaky (after ~3 years of heavy use, not too bad) so rather than putting Smilies on the new lights I took the cue from the 'Cap and built my own version of the un-attached Smilies. But rather than sitting on the platform with the coin, I attached the diffuser to the lens. The new diffuser, which I will call a "Cone Director" to distinguish from the Smilies, is made of a section of a commercial cardboard yarn cone, cut one end to screw into the filter thread on the lens, and the other end to be large enough to just start to vignette. I learned that if you vignette too much you will affect the effective aperture and cause diffraction. Onto the end of the cone I glue a piece of diffusion paper, blocked by black paper at the right diameter to keep the direct reflections of the light off the coin from hitting the lens. Now that I've described it, I realize I need to draw it and take a pic, but that will have to wait...Ray
Edited by rmpsrpms 04/14/2013 9:03 pm
|
|
Moderator
 United States
23522 Posts |
Quote: Thanks Tim! I don't know how to do better. I have a suggestion, but your wallet won't like it: Fullframe. Even greater resolution, yet lower pixel density.
|
|
Pillar of the Community
  United States
4038 Posts |
I've contemplated going FF, but so far have not seen enough benefit. The larger frame will cause more coverage problems, and I would need to increase the magnification to take advantage of the increased frame size and resolution, with the increased mag causing the effective aperture to decrease proportionally. Cents will require M=1.2 instead of M=0.8, so effective aperture decreases 22%. Due to EFSC, the only options I would consider are Canon, likely 5D MkIII. The D800E has a compensated filter arrangement, not removed AA, and no EFSC so would not be a good choice. The 5D-III has a DLA of 10.1, so can work with lenses at M=1.2 of f/4.6. This means it could work with the 75ARD1, so that is very tempting. That said, I'd expect to need to remove the AA filter to do better than I'm doing on the HRT2i.
At some point Canon may introduce a FF camera with 4.3um pixels like the T2i, and maybe truly w/o the AA filter. Doesn't seem all that farfetched. It would have a 47MP sensor! At that point the stars would align and a Cent would require M=1.2, the DLA would be f/6.7, and the min acceptable aperture would be f/3. The 105PN would still fit the bill, and I'd truly see the benefit of FF.
Looking the other way, I've done a lot of work with the XS with great results. DLA is f/9.1, so it can tolerate apertures down to f/5 and still get good sharpness. This allows the 75ARD1 to be used stopped down to f/4.7, a good aperture for it. I've been doing a fair amount of imaging with this configuration.
Edited by rmpsrpms 04/15/2013 01:09 am
|
|
Moderator
 United States
23522 Posts |
Given the results you're getting now, I really can't see a good reason to give up the much greater pixels-per-inch of the APS-C sensor anyway. I think you're resolving above the FF pixel density; visually, the resolution of the original image remains usable viewed at 200%.
|
|
Pillar of the Community
  United States
4038 Posts |
SD...yeah, isn't that cool? That's a result of the removed AA filter. The in-focus areas remain "smooth" even at 400%...Ray
edited to add that the resolution limit for the 105PN is about 3.2um pixel pitch for Cents at M=0.8. This is at f2.8. That makes it useful for the NEX7 with its 3.9um pitch, and the NEX7 has EFSC. I wonder...
Edited by rmpsrpms 04/15/2013 11:36 am
|
|
Moderator
 United States
23522 Posts |
Quote: That makes it useful for the NEX7 with its 3.9um pitch, and the NEX7 has EFSC. I wonder... Oh, dear.  Near as I can tell after a little research, though, there are disadvantages. No remote capture software appears to exist. Although you can output Live View via HDMI, I found evidence that it might not be full resolution, inhibiting the ability to focus effectively. And you'd need EyeFi to automatically export the image. Back to the original image. At a pixel level, there's still room for actual improvement. I mean, it's staggering that I need to reach the individual pixel level to find anything to critique.... All the same, it a magnification sufficient to resolve individual pixels, I see "steps" along straight edges large enough to correct. Moreso along diagonals than verticals/horizontals, of course, and I'm guessing this is probably software-related.
|
|
Pillar of the Community
  United States
4038 Posts |
That was a jpg right out of the camera. Not sure what the camera does during conversion that might make those steps or at least make them worse. I'm also not sure how removing the AA filter really affects the final image. Surely Canon assumes the filter is there, and the camera software does some compensation? I really haven't experimented enough to see how to make the best final image from the HRT2i.
edited to add:
SD, would you like to see a RAW single image to play with?
Edited by rmpsrpms 04/15/2013 8:17 pm
|
| |
Replies: 48 / Views: 6,416 |