| Author |
Replies: 30 / Views: 5,519 |
|
|
|
Pillar of the Community
 United States
891 Posts |
Oh I agree. These guys are the heat. I have picked up a lot of tips to improve my pics. They have come a long way from when I first started. It's the little things that make a pic look better. I would have never thought the pics could have gotten much better.
|
|
Pillar of the Community
 United States
4038 Posts |
Most of the resizing algorithms I've used do a poor job of interpolation, and cause a reduction in sharpness if the resizing ratio is non-integer. I believe the best ratios are factors of 2 since the algorithm only needs to find an average of 4 adjacent pixels (or 16 in case of 4x redux, etc) however this may not be the case for all algorithms. Worst are odd ratios where the algorithm basically has to completely recreate the data. Nothing of the original image survives and sharpness suffers. For instance, below are two photos, one reduced from 1232x816 to 616x408 (50%) and the other to 620x411 (50.325%). The 616x408 is a simple downsizing of the actual photo. The 620x411 is a complete fabrication of the jpg algorithm's imagination. The differences are subtle, but look closely and they become obvious. If your image is too big to display on screen (>50% of the monitor screen width) the image is automatically resized (Bobby published more details on this). SuperDave recommends a 800px screen size, which won't be resized for a 1920px monitor. If you are lazy like me and don't want to crop your images square (or if they are detail shots like below), my only available factor of 2 ratio is 616x408 (8x from my camera's 4928x3264). This is my preferred size, since if I present a 4x reduced image (1232x816) it is further reduced by the display engine to an odd ratio size and sharpness suffers.  
Contact me for photographic equipment or visit my home page at: http://macrocoins.com
Edited by rmpsrpms 08/26/2011 09:57 am
|
|
Moderator
 United States
23522 Posts |
I've always cropped my full-face images as close to perfectly square as I could, before any further processing, to ease the load on sizing algorithms. You're getting absolutely splendid results from an S2 IS, Southern Yankee. As an unabashed Canon fanboi I can only gloat, but in reality it's probably something that many competing brands are capable of in hands as careful as yours. My only advice would be to shoot in as "neutral" a mode as possible, and apply these corrections later in postprocessing. Vivid Mode in an S2 IS adds both sharpening and color saturation in-camera. I believe this to be problematic because the camera uses a very small processor with relatively unsophisticated software, far less capable of seamless results than dedicated postprocessing software on a powerful (by comparison) home PC. The Washington and Lincoln both seem to benefit from this setting, but the Franklin begins to look over-sharpened to me. Unless you're using equipment of such high quality that postprocessing becomes completely irrelevant (like rmpsrpms does), I believe very strongly that a coin photographer needs to be as conversant with certain postprocessing steps as they are with the camera itself. There is nothing wrong with using postprocessing to make your coin look the same on the screen as it does in-hand. Southern Yankee, it's time for you to start considering the tiny steps it will take to make your images, produced on a relatively old and down-spec camera, equal the quality of the very best images posted from sophisticated equipment. You've already stepped into the group of "the best photographers posting at Coin Community," and I think yet more improvement is possible. A consideration, though: not many people are going to be able to see any further improvements you make, simply because they've not seen many images of this quality. 
|
|
Pillar of the Community
 United States
891 Posts |
Wow. Thanks for all the info rmpsrpms And SuperDave. I had no idea about the resizing algorithms. I just cropped and picked a number for resizing and went with it. As for post-processing I don't do any. I always tried to let the camera do all the work. I guess with my limited hardware I will have to start figuring it out. Time to learn the Gimp.
You guys are a big help. Thanks
|
|
Pillar of the Community
 United States
891 Posts |
|
|
Pillar of the Community
 United States
4038 Posts |
I'm curious to hear what are the factors people look for when adjusting photos for various purposes. My philosophy is: - Use full dynamic range, but: > Allow no major are G or B blow outs > Allow lost shadows only around the very edges of the devices - Minimal sharpening Mike, your adjusted image looks too bright to me, especially in Red channel, yet you have lost shadows only in the background. Here is what I would do with the original image in post: 
Contact me for photographic equipment or visit my home page at: http://macrocoins.com
Edited by rmpsrpms 08/28/2011 10:12 am
|
|
Pillar of the Community
 United States
891 Posts |
Ok I understand most of what you are talking about with exception to the G and B blowout. As for the red channel being to bright that is one of the colors I am colorblind to. I don't see all the red you are talking about. That's why I tend to stay away from collecting cents. I will tone down the sharpening some. These were all around 50% sharpening. I'll work on the shadowing. Have to figure out how to just do around the devices.
|
|
Pillar of the Community
 United States
4038 Posts |
What I mean by "blowout" is over-exposure, not just of the overall luminance but of the individual colors in the image. Computer images are a combination of Red, Green and Blue "channels". Most folks just use the overall exposure histogram, but overlook the histograms of individual colors. This is OK for silver coins, but for gold and especially copper the Red channel can be easily blown out. Does the program you are using have histograms available to judge the luminance distribution of the overall image and of the individual colors? Even if you are color blind to the Red in the image, you can use the histogram to graphically make sure it comes out right without having to do it visually.
Contact me for photographic equipment or visit my home page at: http://macrocoins.com
|
|
Pillar of the Community
 United States
4038 Posts |
Forgot to mention that your lighting is excellent and gives good shadow definition around the devices already without having too many highlights. The adjustment I made was to contrast to push the darkest shadows just into black. When I do this on your image, the shadows are nearly continuous around the devices and look great. The program I use (Nikon ViewNX) has functions called "show lost shadows" and "show lost highlights" that I use extensively. Not sure if other programs have this available.
|
|
Pillar of the Community
 United States
4038 Posts |
Here is your original image with brightness cranked up to show where the lost shadows are. Looks very good, with shadows just showing up nicely around the devices. This shows you have the lights high enough to define the devices but not so high as to make the shadow areas too prominent. It's an excellent result.  PS: Also increased sharpness of the image to amplify the local contrast. In reality your original image had no lost shadows, they were all just above black so I used sharpening to push them over the edge to show the effect.
Contact me for photographic equipment or visit my home page at: http://macrocoins.com
Edited by rmpsrpms 08/28/2011 2:58 pm
|
|
Pillar of the Community
 United States
891 Posts |
Thanks for that great explanation. I misunderstood about what you said about the shadows, but now with reading about it on the net and your illustration it's starting to make sense. Have to look into what my program has on that. My program has individual histograms on the colors but I never looked into using those. Always just used the overall histogram. So much to take in to account for just a simple pic. I see this is going to be quite a learning experience. Thank you so much for your help.
|
|
Pillar of the Community
 United States
4038 Posts |
Folks have complained about not being able to white balance the IKEA Jansjo LEDs. In the process of trying to find a simple way to improve the illumination uniformity of these lights I may have stumbled upon the reason for white balance problem and a way to solve it... A simple tube reflector seems to work well for concentrating the beam from the Jansjo, increasing the illumination intensity and uniformity. While documenting this, I found that the Jansjo has significant yellow and red color fringes, probably due to the design of the focusing lens. Turns out the tube reflector also helps tame these fringes! Below are pictures of two IKEA lights shining on white index cards. White balance was done on the central focus of the beam. The first picture shows the two lights in "bare" configuration. The second picture shows the two lights again, with one having a simple white Opalux tube mounted on it. Notice the brighter,more controlled and uniform illumination. The third picture is same as the second but with brightness significantly reduced so you can better see the uniformity, and this is where it gets interesting. The "bare" Jansjo is showing significant color fringing with yellow moving into red shifts at the edge of its beam. The tube reflector eliminates the red and reduces the yellow shifts while simultaneously increasing brightness. A simple modification like this may help folks improve the white balance of their Jansjo's without having to add diffusers and lose illumination intensity and beam concentration (and ability to emphasize luster). There is still work to be done further improving the color and illumination uniformity, but first step is always recognizing the problem... Here are the pics: Pic1: "Bare" Jansjos  Pic2: with tube reflector on one Jansjo  Pic2 with reduced brightness to show illumination pattern 
Contact me for photographic equipment or visit my home page at: http://macrocoins.com
|
|
Moderator
 United States
23522 Posts |
To add one small point to rmpsrpms' excellent explanation: The lighted area you're shooting for the Custom White Balance shot must fill the whole sensor. Whatever that takes; even moving the camera closer to the shot than you'll actually use for the coins. If any of the fringe shows on the sensor, it'll affect white balance.
|
|
Pillar of the Community
United States
509 Posts |
Ray, I've also noticed the wide beam of the Jansjo LED's. Your tube reflector looks like it might have interesting possibilities. I've found my better results seem to come using reflected light off of the aluminum reflector mounted overhead between the camera body and lens. It will be interesting to play with. Not many problems though with WB anymore thanks to you and SD. I'm still looking for that "ultimate" shot. Can't be done I know but sure is enjoyable chasing after it.
|
|
Pillar of the Community
 United States
4038 Posts |
Here's another try at it. This time using a short section of a yarn cone, with white index card pressed inside. This equalizes the uniformity within the beam and increases intensity further. As before, I'll show the "normal" exposure look vs the stock Jansjo and then a reduced brightness look to show uniformity better. I reduced the brightness even further than earlier shots to show bare Jansjo uniformity around the same spot size as the modified Jansjo. Later tonight I'll take some pictures with them and see if there is any benefit in coin photography...  
|
| |
Replies: 30 / Views: 5,519 |