Effects of Photo Compression on Critiques

[Don Schaeffer]

I just had the thought that we are critiquing factors like sharpness and noise levels in photos that have been compressed by a factor of 4 or more in order to be acceptable by the server at the forum. How valid are those critiques? I have found that the compression process changes the level of visible detail in my photos, even changes contrast as well as sharpness. Anyone care to comment?

[Mr Yuck]

I think they are valid, what I do to avoid these is just post the photo in the gallery or on an external site, and then link a larger version. (on the medium, forum sized version)

kinda like this:

[URL="URL to Large Image"][IMG]URL to small Image[IMG][URL]

*with a / at the terminating IMG and URL tags.

[GB1]

Quite valid. From what I can see, over-sharpening creates a very grainy look that seems artificial. This has a negative effect on the way the image presents, which is what photography is pretty much all about.

I believe that sharpening is an algorithm run on the image which increases contrast between adjacent colors to create the illusion of sharpness. You might notice that night shots seem very sharp - the contrast between the lit and dark areas is usually quite high.

Jpeg and GIF compression are both "lossy", meaning that you lose image quality and/or color space when you compress. It's sometimes needed to shrink the image to a manageable size, but only very little compression is needed here - PR's 640 pixel limitation pretty much limits the practical size of an uncompressed image to ~ 300 kb, depending on the image. Since PR's upload limit is 244 kb, you only need a very small amount of compression....With Corel, you can control the amount (0-100%), but not sure you have much control with PS (which is surprising since PS is considered the 'flagship' of digital editing software).

[readingr]

Sharpening is a skill especially if the size of the image is to be reduced.

How I do it.

I edit the file using PS and save as a PSD.

I do not sharpen the edited picture before resizing. It is always the last thing I do before printing or saving as a web image. If you do it now then you will exacerbate the halos produced during sharpening and resizing.

1. Add the copyright
2. I use a macro which resizes in 25% increments down to whatever size I want.
3. Once re-sized I use the unsharp mask and will be very careful about how much I apply.
4. Save as JPEG with no or very little compression and usually the file comes in just under the limit.

I use PSP to add the copyright and run the macro.

Roger

Roger

[freygr]

Before I resize a photo smaller, I blur the photo. This helps as the resize acts like applying a sharpening filter. The program I use will calculate the file size so I can adjust the default compression settings and the per cent compression to get the best JPG compression possible for image quality for posting.

[Don Schaeffer]

Very interesting suggestions here.

[livin4lax09]

host off-site. imageshack.us, imagesocket.com, imgplace.com, putfile, photobucket, tinypic, there are a ton of sites. You can even upload it to your gallery then link it directly from there by right clicking and copying the location of the image. There's no reason to compromise the quality of an image in order to post it. I usually save at 10 and don't have any problem uploading when I want to.

[livin4lax09]

roger, I found your technique to be interesting so I decided to give it a try... what i found is exactly the opposite of what I expected. using your technique except upsizing works exceptionally well, but downsizing? not so much.

action created to resize to 90% of original size, bicubic 72 ppi, then run until reached desired size. USM of 100, .3, 0, then saved as 12 JPEG


resized straight to 600 pix high bicubic, 72 ppi, USM 100, .3, 0,


resized straight to 600 pix high, bicubic, 72 ppi, USM 200, .3, 0


no other work was done to these images, simply opened, resized, sharpened, converted to sRGB, and saved.

is the final image a bit too sharp? Yeah probably, but I think this shows us photoshop has a pretty good handle on their resizing algorithms, and manages to do it in the least-destructive fashion, while allowing you to get a pretty good result in the end.

[livin4lax09]

Quite valid. From what I can see, over-sharpening creates a very grainy look that seems artificial. This has a negative effect on the way the image presents, which is what photography is pretty much all about.

I believe that sharpening is an algorithm run on the image which increases contrast between adjacent colors to create the illusion of sharpness. You might notice that night shots seem very sharp - the contrast between the lit and dark areas is usually quite high.

Jpeg and GIF compression are both "lossy", meaning that you lose image quality and/or color space when you compress. It's sometimes needed to shrink the image to a manageable size, but only very little compression is needed here - PR's 640 pixel limitation pretty much limits the practical size of an uncompressed image to ~ 300 kb, depending on the image. Since PR's upload limit is 244 kb, you only need a very small amount of compression....With Corel, you can control the amount (0-100%), but not sure you have much control with PS (which is surprising since PS is considered the 'flagship' of digital editing software).

well you have values to choose from, 1-12. I find 10 almost always brings it in under the limit, with barely any visible difference in quality.

[readingr]

roger, I found your technique to be interesting so I decided to give it a try... what i found is exactly the opposite of what I expected. using your technique except upsizing works exceptionally well, but downsizing? not so much.

is the final image a bit too sharp? Yeah probably, but I think this shows us photoshop has a pretty good handle on their resizing algorithms, and manages to do it in the least-destructive fashion, while allowing you to get a pretty good result in the end.

Brent,

You missed the part where I use PSP (Paint Shop Pro) with that method for resizing. If I use PS to downsize I just use the image resize - PS seems to be better at downsizing that PSP. Only recently started to use PS for watermarking and resizing.

Roger

[berrywise]

If they look a little to sharp just squint at the image that always seems to do the trick

I'm lazy, I resize my image, sharpen it using smart sharpen, then use the save for the web function in Photoshop.

[photophorous]

I just had the thought that we are critiquing factors like sharpness and noise levels in photos that have been compressed by a factor of 4 or more in order to be acceptable by the server at the forum. How valid are those critiques? I have found that the compression process changes the level of visible detail in my photos, even changes contrast as well as sharpness. Anyone care to comment?

I think that critiquing those aspects of the photo that are effected by resizing is valid, because most photographers these days present their photos on the web. Most of the time, that's not our intended final product, but in some cases it is. When someone tells you the contrast, noise, sharpness, etc. looks bad, it's good information to know whether it's a result of post processing or your camera technique. You can then go back and check the full size file to determine where the problem was introduced.

It seems like there are endless ways to sharpen photos, and everyone has their own method. For me, it depends a lot on the type of shot. If I have something like a portrait, that doesn't have a lot of fine detail, I'll run noise reduction first, which smooths out the skin a little (usually a good thing), and then do selective sharpening after resizing. For a very detailed shot, like a landscape, I usually do a small amount of sharpening before I resize, and then again after I resize. I find that sharpening before the resize helps preserve the fine details that get smoothed out during the resize. Too much sharpening is bad whether it's done before or after. In the case of a landscape, if noise is an issue, you can use masking layers to exclude areas where the noise is most visible, like the sky.

Paul

[another view]

I think that critiquing those aspects of the photo that are effected by resizing is valid, because most photographers these days present their photos on the web. Most of the time, that's not our intended final product, but in some cases it is.

I think you're right. However, I think that if you were to compare any on-screen version of an image to a print there's a bigger problem - color management. If I upload an image that looks good on my calibrated monitor, then it looks bad (or at least different) on the el-cheapo tube at work or my wife's laptop that I'm using at the moment.

Paul's point is very valid, and I don't think there's any way to solve the problem of color management. It's probably the same reason that people might return stuff bought online because the color wasn't right. Maybe they're upset that the item didn't look anything like the picture - but in reality it was their monitor that was the problem...

[GB1]

well you have values to choose from, 1-12. I find 10 almost always brings it in under the limit, with barely any visible difference in quality.

Brent - Ooops, you're right, what I meant was the sharpening tool in PS, not the compression. Is there any way to control the percentage of sharpening in PS? I haven't seen one yet.

Gb

[livin4lax09]

yeah, its a more advanced tool though. The unsharp mask is what is commonly used. You have to put in amount, radius, and threshold, but it gives you a much better control of the sharpening.

Roger, sorry I missed that. Now I understand what PSP means. Thanks for clearing that up.

[berrywise]

Ran into this tonight:

http://tutorials.watchandlearnphotos...c_sharper.html