Tweet
By DJ Dubovsky on Wednesday, August 08, 2001 - 12:32 pm:
OK, I have asked this question before to other forums and never got a response so I'm assuming it's a stupid question or no one has ever had to deal with this but me.
Have you ever over adjusted a photo and you notice it's caused posturization. I think that's what it's called. It's when the shading in areas where shadows and midtones are seem no longer smooth and gradual but with definate lines and dots of seperation? If you try to blur it out or use dust and scratches to clear it up, it just makes the dark and light areas more outlined.
There's all kinds of advice on how to avoid this but no one really tells you how to deal with it if it already exists. Has anyone found a neat way to clean those areas up? I'm not even sure if I'm explaining it right.
Well, that's my stupid question for the day. Thanks for your help.
DJ
By Ed Ladendorf on Wednesday, August 08, 2001 - 01:16 pm:
Hi Debbie,
Just remember that the guy speaking is not one of the best informed on the site. But if you have a chance to rescan the photo, and are able to scan in high bit mode, you might be able to do more tonal correction without resulting in posterization. Recently, I have been in a mind of "find out what you can about high bit scanning". I have had *many* answers about the benefits of high bit scanning. A good number of them say that there is no benefit in high bit scanning, while the other school of thought is that there is, at least in some circumstances, a great deal of benefit. If I had something like you describe, I would scan in high bit mode, make tonal corrections, then convert it to 8 bit mode, just to see if there is much of a difference. My own opinion is that there would be more of a chance for making the tonal corrections successfully than there would be on an 8 bit scan. Hope this helps, even if it is not an "after the fact fix".
Ed
By kathleen on Wednesday, August 08, 2001 - 01:40 pm:
dj,
badly want to be able to help you, nothing in my own personal data bank.
i did find this, which helped me understand the nature of the problem, which it sounds like you understand already:
Imagine that your histogram is a Slinky® viewed from the side, and that when you first adjust Levels (or Curves) you stretch that spring out toward either end of the continuum. Now you have a histogram that resembles a comb viewed from the side, and the spaces in between the teeth of that comb are areas where there is no native tonal information...you've "cavitated" your histogram. Your software will fill those spaces, but the process of doing so isn't particularly intelligent (Refer to Interpolation & Extrapolation). In fact, in extreme cases, the digital compensation may produce noticeable posterization
and it sounds like this is not a problem you created, but was pre-existing, right?
the only other thing that came to my mind was, that posterization filter, if you ran that and looked at what it did to the histogram, maybe it would help know which way to go (backwards thinking, my specialty) and i guess knowing how you want the histogram to look and making it look that way are two different things.
onward and upward : ) good luck k
By thomasgeorge on Wednesday, August 08, 2001 - 02:40 pm:
DJ, There's no such thing as a stupid question, only stupid answers. That being said I shall attempt to prove it! You could try duplicating the posterized layer, use gaussian blur at a light setting and repeat until the tones blend. Then use the history brush (experiment with the softness,fade etc.) to paint back the most important details.When flattening try different blending modes also. However the best way is probably to use the revert command or its twin in whatever processing software you are using. Good luck, Tom
By DJ Dubovsky on Wednesday, August 08, 2001 - 05:36 pm:
Thanks guys. Your responses are all helpful. I can scan to avoid these problems but sometimes I have to deal with what I get. Mostly it's just a small area but aggrievating non the less.
Tom - You had me laughing after reading the first 2 lines. But I will definately give your suggestions a try. I haven't tried those ideas so maybe they just might work.
Kathleen - I went to that website and I think it will give me alot of valuable info. I love links to good tutorials and tips. I will definately check it out more in depth. Thanks
ED - I've read your posts to problems on this site several times and I think you are alot better than you are willing to admit. You're work on your wife's photo proves that. Besides as I said the pro sites ignored my question all together. At least you guys were eager to help. Thanks again
DJ
By Ed Ladendorf on Wednesday, August 08, 2001 - 06:19 pm:
I appreciate your comments Debbie, but I just don't want people to think that I am a pro when I'm much closer to being a beginner, or that my suggestions will necessarily work. I just might *think* something will work, when it actually won't in practice. But I'll admit that I rather enjoy helping others if I can. Thanks.
Ed
OK, I have asked this question before to other forums and never got a response so I'm assuming it's a stupid question or no one has ever had to deal with this but me.
Have you ever over adjusted a photo and you notice it's caused posturization. I think that's what it's called. It's when the shading in areas where shadows and midtones are seem no longer smooth and gradual but with definate lines and dots of seperation? If you try to blur it out or use dust and scratches to clear it up, it just makes the dark and light areas more outlined.
There's all kinds of advice on how to avoid this but no one really tells you how to deal with it if it already exists. Has anyone found a neat way to clean those areas up? I'm not even sure if I'm explaining it right.
Well, that's my stupid question for the day. Thanks for your help.
DJ
By Ed Ladendorf on Wednesday, August 08, 2001 - 01:16 pm:
Hi Debbie,
Just remember that the guy speaking is not one of the best informed on the site. But if you have a chance to rescan the photo, and are able to scan in high bit mode, you might be able to do more tonal correction without resulting in posterization. Recently, I have been in a mind of "find out what you can about high bit scanning". I have had *many* answers about the benefits of high bit scanning. A good number of them say that there is no benefit in high bit scanning, while the other school of thought is that there is, at least in some circumstances, a great deal of benefit. If I had something like you describe, I would scan in high bit mode, make tonal corrections, then convert it to 8 bit mode, just to see if there is much of a difference. My own opinion is that there would be more of a chance for making the tonal corrections successfully than there would be on an 8 bit scan. Hope this helps, even if it is not an "after the fact fix".
Ed
By kathleen on Wednesday, August 08, 2001 - 01:40 pm:
dj,
badly want to be able to help you, nothing in my own personal data bank.
i did find this, which helped me understand the nature of the problem, which it sounds like you understand already:
Imagine that your histogram is a Slinky® viewed from the side, and that when you first adjust Levels (or Curves) you stretch that spring out toward either end of the continuum. Now you have a histogram that resembles a comb viewed from the side, and the spaces in between the teeth of that comb are areas where there is no native tonal information...you've "cavitated" your histogram. Your software will fill those spaces, but the process of doing so isn't particularly intelligent (Refer to Interpolation & Extrapolation). In fact, in extreme cases, the digital compensation may produce noticeable posterization
and it sounds like this is not a problem you created, but was pre-existing, right?
the only other thing that came to my mind was, that posterization filter, if you ran that and looked at what it did to the histogram, maybe it would help know which way to go (backwards thinking, my specialty) and i guess knowing how you want the histogram to look and making it look that way are two different things.
onward and upward : ) good luck k
By thomasgeorge on Wednesday, August 08, 2001 - 02:40 pm:
DJ, There's no such thing as a stupid question, only stupid answers. That being said I shall attempt to prove it! You could try duplicating the posterized layer, use gaussian blur at a light setting and repeat until the tones blend. Then use the history brush (experiment with the softness,fade etc.) to paint back the most important details.When flattening try different blending modes also. However the best way is probably to use the revert command or its twin in whatever processing software you are using. Good luck, Tom
By DJ Dubovsky on Wednesday, August 08, 2001 - 05:36 pm:
Thanks guys. Your responses are all helpful. I can scan to avoid these problems but sometimes I have to deal with what I get. Mostly it's just a small area but aggrievating non the less.
Tom - You had me laughing after reading the first 2 lines. But I will definately give your suggestions a try. I haven't tried those ideas so maybe they just might work.
Kathleen - I went to that website and I think it will give me alot of valuable info. I love links to good tutorials and tips. I will definately check it out more in depth. Thanks
ED - I've read your posts to problems on this site several times and I think you are alot better than you are willing to admit. You're work on your wife's photo proves that. Besides as I said the pro sites ignored my question all together. At least you guys were eager to help. Thanks again
DJ
By Ed Ladendorf on Wednesday, August 08, 2001 - 06:19 pm:
I appreciate your comments Debbie, but I just don't want people to think that I am a pro when I'm much closer to being a beginner, or that my suggestions will necessarily work. I just might *think* something will work, when it actually won't in practice. But I'll admit that I rather enjoy helping others if I can. Thanks.
Ed