Increasing the bandwidth requirements for some slows down the performance for all. It's not quite as easy as that.
I contemplated the "pay for extra quality" too, actually. For the most part I'm fine with smugmug's compression quality, but I have a handful of photos that due to their nature just artifact like crazy. It would be nice to just have an "edit photo" button that would refresh the thumbnail with some sort of "delicate image" considerations taken in...
But I think that by-and-large, simply upping the qualiy of every single S, M & L image on one's website would greatly increase lag-time. All you DSL etc. users have NO idea how slow these images are loading for the 56k-ers out there. My girlfriend just recently visited her granparents, and she wanted to show them my photos but they basically had to stop after a while because I just have so many photos that it would have taken literally all day and all night. Now with my DSL connection I can show someone practially my whole portfolio / exploits galleries in 30-45 minutes. I know more and more people are converting to faster internet connections, but I think it's wise that smugmug is not just leaving the 56k-ers in the dust. If a McAskill (sp?) wants to chime in and give us some stats on the ratio of cable-to-dialup users on Smugug, I bet we'd be surprised.
I never thought I'd find myself saying this, but other than maybe, maybe, MAYBE adding a single image "refresh thumbnail" button, we definitely must leave image compression quality the way it is right now.
If you REALLY want higher quality images, then use a Photoshop action to take your high-res photos and make them 700-800 pixels, run your best shot of USM, and upload those photos into your original gallery. Disable printing for this gallery. Move the high-res photos to a new, "all thumbs" gallery, disable larges and title the gallery "such-and-such event/genre PRINTS gallery" and post a link to it in the original gallery. In my opinion that is a very professional means of displaying your photos, and if you want, it wouldn't hurt to explain to people why you've set things up this way. I bet you'd get lots of comments from people saying that your 800 pixel "high quality thumbnails" load very slowly!
I understand what you are saying. But this is why I think we should have a choice. Because I couldn't care less about those that have 56K internet connections. Every single friend and family member that I know, all have DSL or Cable. I'm not trying to run a professional gallery for customers. I just want to see my photos clean and sharp without artifacts in them. Why is this such a hard thing to ask?
Choice, choice, choice. Customers like choices. Something Smugmug does a terrific job at, with exception to this issue.
I understand what you are saying. But this is why I think we should have a choice. Because I couldn't care less about those that have 56K internet connections. Every single friend and family member that I know, all have DSL or Cable. I'm not trying to run a professional gallery for customers. I just want to see my photos clean and sharp without artifacts in them. Why is this such a hard thing to ask?
Choice, choice, choice. Customers like choices. Something Smugmug does a terrific job at, with exception to this issue.
While I can see the artifacts in my resized images, I don't particularly mind. But I don't think that my not minding is a good reason to tell you not to mind. If indeed everyone is as happy with their jpeg compression as they say they are, then adding the capability to allow certain pros to increase it could possibly even be amortized across the entire range of smugmug customers and the change would be nearly imperceptible (a bit like how the "infinite" storage gets amortized out.) A pro acknowledges that decreasing compression increases size (who doesn't?) and would take longer to download. But they are grown adults and know who their audience is, and will generally make an informed decision about load times if they value image quality enough to want to compromise them.
I don't know how smugmug maintains its huge database of photos, but I can't think of any scheme by which this would be a difficult change, technically. The argument against it, that I've heard, is that people will go complaining to smugmug if the load times are too slow. I wouldn't think so, since most people will think they're visiting the photographer's site, not the photographer's site hosted on smugmug. They will tell whomever they think is the owner of the site (in this case the pro photographer) that they think their site is too slow. If the pro gets too many of these emails, he/she might consider increasing the compression again. But allowing this change doesn't seem like such a big deal.
All this arguing against the change seems, in some cases, inspired by cognitive dissonance... but I may just be reading things into it that I shouldn't
But that also means that people can take those images and print them themselves. The other option is to upload a <800px image and then you necessarily lose many of the benefits of a pro account (mostly being able to sell photos online for a profit.) Or the third option is to just accept the artifacts, which doesn't seem like a very good one either.
Isn't it like buying a new car and finding a scratch on it? You know it won't affect the car's performance in the least, but it still bugs you if you know about it and see it every time you use the car.
ok - smugmug allows you the choice of showing your original images
Andy let's not get into that one again :. The interface for viewing your original images is obviously put in place to discourage just that, since in order to view the original one must click on the link and open a new window, then close the window before selecting the next thumb only to repeat the whole process again. VERY inconvenient.
The whole point of wanting crisp photos is so that people browsing my galleries will see my photos the way I meant to have them seen. I didn't spend hours post-processing my photos, getting them just the right sharpness, sizing to 1024, etc only to have them viewed all jagged and fugly. I really don't think this is too much to ask. Like I said, I would be willing to pay a premium for this.
By the way, I completely agree with what pumpkin has said about the affect of allowing this option to certain pro users being amortized across their entire user base.
Choice, choice, choice. Customers like choices. Something Smugmug does a terrific job at, with exception to this issue.
Actually, the problem is really much simpler than this. The problem is that smugmug uses a constant quantization matrix (i.e. quality setting) for making the thumbnails. Now the smugmug guys are real cool and all, but for people as tech savy as they are this really is patently stupid. (To be fair, they are really really busy putting out all sorts of cool customization features so cutting some slack is in order :).
What they should be doing if they are interested in good image quality and fast downloads is targeting a constant thumbnail file *size*. This is what pretty much every digital camera on the market does. Take a picture of static on your TV or a clear blue sky and the file sizes come our remarkably similar in size. This of course seems impossible since there are vastly different amounts of information in each image, until you look at the JPEG headers and you'll discover the quatization matrix for each image is drastically different - the blue sky less lossy (less rounding in the quantization matrix) and the static more lossy (more rounding).
So the general idea is to adjust the quantization matix (quality setting) until you get the desired size. This results in much nicer looking photos. Why? Well, in very complicated images that require lower quality settings to meet the target file size the artifacts are less visible - the complicated scene information hides them. In images with large uniform patches at the lower quality settings artifacts would show up like mad at the edges (sky for example) - but large uniform patches compress real well so the file can have a much higher quailty setting and still meet the target file size.
So far most of the example images posted in threads like these by people who feel their photos were mauled by the compression have had large uniform areas in them (sky) and the artifacts are all visible in the sky. And if you look at the file *sizes* they are way smaller than they need to be. Here's an example from one of the few non-private galleries I've got up at the moment:
There are some scary artifacts around the lightning. This image has lots of uniform space and had noise reduction applied to it, as a result it compresses well. The large image is only 30KB!!!! Unfortunately from the clouds to the bolt is a big step function and artifacts show up real nice in that area. A higher quality setting would remove the artifacts and still keep the image at a managable size.
Now this image is 70KB, more than *twice* the size of the first!!! It's contrasty and there are few uniform areas. You can pick up a little artifacting where the Sierra's meet the sky, but its not objectionable, and everywhere else in the image the artifacts are hidden by other scene information (e.g. the top of the dunes is a very sharp edge, but the artifacting is much harder to see than in the lightning image).
Long story short, the current constant quality settings are suboptimal for both image quality and download time. The quality should be set adaptively to achive a target file size for the thumbnails. This is a back end change for the smugmug folks (we love you, by the way, it's just tough love ) and wouldn't require any messy user interface changes. It would probably make the fanatical pixel peepers a lot happier about the vast majority of the images they don't like and would keep the 56Ker's out in their mud huts using smoke signals to connect to the internet happy as well.
Anywho, this has come up before, I think even Baldy mentioned they were looking into it. I sure hope they do. It will result in both good bandwidth performance and image quality. It can't be that hard to do if your camera is doing it at 5 fps. I think the problem is many of the command line processing tools out there don't have a target file size option so they might need to wrap a script around whatever they are using to do it right, and as we know it isn't like they're sitting on their duffs drinking beer at the moment.
Oh, and I think I mentioned this on another thread, if you are truely disturbed by what are in fact pretty minor artifacts I strongly suggest you don't visit your clients or family and friends who are viewing your works of art on crappy uncalibrated monitors. The drastic color shifts are *way* more objectionable.
Actually, the problem is really much simpler than this. The problem is that smugmug uses a constant quantization matrix (i.e. quality setting) for making the thumbnails. Now the smugmug guys are real cool and all, but for people as tech savy as they are this really is patently stupid. (To be fair, they are really really busy putting out all sorts of cool customization features so cutting some slack is in order :).
What they should be doing if they are interested in good image quality and fast downloads is targeting a constant thumbnail file *size*. This is what pretty much every digital camera on the market does. Take a picture of static on your TV or a clear blue sky and the file sizes come our remarkably similar in size. This of course seems impossible since there are vastly different amounts of information in each image, until you look at the JPEG headers and you'll discover the quatization matrix for each image is drastically different - the blue sky less lossy (less rounding in the quantization matrix) and the static more lossy (more rounding).
So the general idea is to adjust the quantization matix (quality setting) until you get the desired size. This results in much nicer looking photos. Why? Well, in very complicated images that require lower quality settings to meet the target file size the artifacts are less visible - the complicated scene information hides them. In images with large uniform patches at the lower quality settings artifacts would show up like mad at the edges (sky for example) - but large uniform patches compress real well so the file can have a much higher quailty setting and still meet the target file size.
So far most of the example images posted in threads like these by people who feel their photos were mauled by the compression have had large uniform areas in them (sky) and the artifacts are all visible in the sky. And if you look at the file *sizes* they are way smaller than they need to be. Here's an example from one of the few non-private galleries I've got up at the moment:
There are some scary artifacts around the lightning. This image has lots of uniform space and had noise reduction applied to it, as a result it compresses well. The large image is only 30KB!!!! Unfortunately from the clouds to the bolt is a big step function and artifacts show up real nice in that area. A higher quality setting would remove the artifacts and still keep the image at a managable size.
Now this image is 70KB, more than *twice* the size of the first!!! It's contrasty and there are few uniform areas. You can pick up a little artifacting where the Sierra's meet the sky, but its not objectionable, and everywhere else in the image the artifacts are hidden by other scene information (e.g. the top of the dunes is a very sharp edge, but the artifacting is much harder to see than in the lightning image).
Long story short, the current constant quality settings are suboptimal for both image quality and download time. The quality should be set adaptively to achive a target file size for the thumbnails. This is a back end change for the smugmug folks (we love you, by the way, it's just tough love ) and wouldn't require any messy user interface changes. It would probably make the fanatical pixel peepers a lot happier about the vast majority of the images they don't like and would keep the 56Ker's out in their mud huts using smoke signals to connect to the internet happy as well.
Anywho, this has come up before, I think even Baldy mentioned they were looking into it. I sure hope they do. It will result in both good bandwidth performance and image quality. It can't be that hard to do if your camera is doing it at 5 fps. I think the problem is many of the command line processing tools out there don't have a target file size option so they might need to wrap a script around whatever they are using to do it right, and as we know it isn't like they're sitting on their duffs drinking beer at the moment.
Oh, and I think I mentioned this on another thread, if you are truely disturbed by what are in fact pretty minor artifacts I strongly suggest you don't visit your clients or family and friends who are viewing your works of art on crappy uncalibrated monitors. The drastic color shifts are *way* more objectionable.
this discussion is fascinating...really... i'm looking forward to your answers
three versions of the same shot. all from a canon 1Ds Mark II shot, the full-size jpg is just under 8 megabytes.
which one was made by smugmug, and which was made by me, and finally, which one was made by me and also compressed by half?
The last one was made by smugmug, the first one was compressed in half, and the middle one was made by you. But it took a good long time of nose to the screen staring to get it.
It's the pixelation around the necklace/sky and shirt/water border that gives it away.
Y'all don't want to hear me, you just want to dance.
The last one was made by smugmug, the first one was compressed in half, and the middle one was made by you. But it took a good long time of nose to the screen staring to get it.
It's the pixelation around the necklace/sky and shirt/water border that gives it away.
Actually it doesn't take that much nose to the screen to see the differences. The middle one is far sharper than either of the other two. And the last one has lost so much detail that the texture of the girls skin is completely gone. It almost looks like it is slightly out of focus.
And for another test, how about this. Notice how in one of the photos, the girls face almost looks blotchy because of the pixelation. The other is crystal clear and sharp. Notice I am not even mentioning the pixelation between the high contrast areas. I am speaking of the model's face texture directly.
And for another test, how about this. Notice how in one of the photos, the girls face almost looks blotchy because of the pixelation. The other is crystal clear and sharp. Notice I am not even mentioning the pixelation between the high contrast areas. I am speaking of the model's face texture directly.
why are you showing -M? i mean, c'mon - with most any camera today, you're talking a huge amount of difference between a -M size and the original
Actually it doesn't take that much nose to the screen to see the differences. The middle one is far sharper than either of the other two. And the last one has lost so much detail that the texture of the girls skin is completely gone. It almost looks like it is slightly out of focus.
ahh now you get to a whole 'nother can of tomatoes: sharpening! more later
The point here was to show you the difference between a properly re-sized and compressed image versus how Smugmug does it. Hence the M size. But for your satisfaction:
unsavory, what's your workflow? i'd also like to see a shot that you process, full size, and upload to smugmug and then link the -L and also a home-brewed 800px version. thanks!
unsavory, what's your workflow? i'd also like to see a shot that you process, full size, and upload to smugmug and then link the -L and also a home-brewed 800px version. thanks!
Will do Andy. Unfortunately I am running out to a birthday party right now. So I'll have to pick this up tomorrow. Thanks for taking the time.
The point here was to show you the difference between a properly re-sized and compressed image versus how Smugmug does it. Hence the M size. But for your satisfaction:
I've got to tell you, I really have to look hard to see a difference in the L photos. Do you really think it'd be a issue if a potential customer saw the smugmug compressed version? I know I'm coming off as a part of "them" a little bit here, but really, honestly (and maybe it's just me, maybe my eye isn't dicerning enough) I have to get right up to the screen and switch back and forth between pictures to see if I can tell if any pixels change places.
Am I just not being ciritical enough?
Y'all don't want to hear me, you just want to dance.
a big part of this whole issue is that photographers will always notice this stuff more than buyers
That's true and that was a tough thing for me to come to terms with. Of course, maybe it shouldn't have been that tough because I don't sell much of anything in the first place .
Y'all don't want to hear me, you just want to dance.
this discussion is fascinating...really... i'm looking forward to your answers
three versions of the same shot. all from a canon 1Ds Mark II shot, the full-size jpg is just under 8 megabytes.
which one was made by smugmug, and which was made by me, and finally, which one was made by me and also compressed by half?
(no cheating by looking at the file size!)
resized to 800x533 and compressed to <100kb by me in photoshop:
resized by me in photoshop to 800x533, no compression:
compressed by smugmug to 800x533
so, i've put the answers here. mike you were spot on - and so here are my comments.
the middle shot, to me, seems a bit oversharp - as i've matured as a photographer i feel that not everything needs to cut your skin sharp! some stuff, yeah. i actually prefer the look of the third version, and to unsavory: there are no details lost - this is the original file as i processed it from the 16mpx canon 1Ds Mark II... anyhow, the sharpening thing is such a personal preference that we shouldn't debate that.
my point, that i'm attempting to make, is that i feel that the compression is acceptable for the -L images that i show in my galleries. could it be less compression? sure - and the quality would even go up - but we cannot ignore the tradeoff of file size vs. viewer experience (speed).
BaldyRegistered Users, Super ModeratorsPosts: 2,853moderator
edited August 27, 2005
I have the mother of all headaches so won't type much, but that post of kwalsh's is exactly what I was looking for. It makes a lot of sense.
One thing that confuses me is I'm looking at a folder of differnet shots from my 20D, all at the same medium compression setting, and they range from 606k to 1518k depending on the scene I understood kwalsh to say virtually every digital camera puts out a constant size image (he said thumbnail, but we're talking more than thumbs, correct?).
Oh, and I think I mentioned this on another thread, if you are truely disturbed by what are in fact pretty minor artifacts I strongly suggest you don't visit your clients or family and friends who are viewing your works of art on crappy uncalibrated monitors. The drastic color shifts are *way* more objectionable.
Ken
ken, a great post with some excellent food for thought. and your final point is soooooo true - i have seen some folks with multi-thou$and dollar displays and they look like the dog barfed on the screen it's so green. :puke
One thing that confuses me is I'm looking at a folder of differnet shots from my 20D, all at the same medium compression setting, and they range from 606k to 1518k depending on the scene I understood kwalsh to say virtually every digital camera puts out a constant size image (he said thumbnail, but we're talking more than thumbs, correct?).
*NEWS FLASH FOR KEN*
Beware small number statistics for they shall make you look like a moron! :crazy
So, yeah Baldy you're right on that one. "Virtually every" was strong language for me to use on a survey of 5 cameras. Turns out none of them were from Canon. Canon uses a constant quantization matrix. So, for a larger survey across more manufacturers and cameras:
So, it would appear that for any given manufacturer they've either set compression dynamically or constant across their product line more or less since the dawn of time (there are some old cameras in that list). Dynamic seems to be the more common, but by no means "virtually every". Of course if you take your five samples from Nikon, Sony and Olympus it'll look that way .
What do you guys use to convert on the back end (if you're comfortable saying)? If it's freeware I'd love to try out some simple scripts to see how easy/hard a target file size scheme is on my own system.
Ken
P.S. Hope your headache is better.
0
Options
BaldyRegistered Users, Super ModeratorsPosts: 2,853moderator
edited August 28, 2005
Y'know, yesterday my head felt like it had been run over by a truck. But I'm good today.
We use Image Magick. I talked to Onethumb last night and I think we're pretty open to making the small images (byte-wise) bigger — the ones with the broad areas of solid color — if there's a way to do it without also making the fatty images that don't show artifacts even fatter, sounds like goodness.
Y'know, yesterday my head felt like it had been run over by a truck. But I'm good today.
We use Image Magick. I talked to Onethumb last night and I think we're pretty open to making the small images (byte-wise) bigger — the ones with the broad areas of solid color — if there's a way to do it without also making the fatty images that don't show artifacts even fatter, sounds like goodness.
Oh my god, OUCH!!!! I'll stay on topic and leave best wishes in the other thread. I'll try some stuff out with ImageMagick in a python script and see how it goes.
We use Image Magick. I talked to Onethumb last night and I think we're pretty open to making the small images (byte-wise) bigger — the ones with the broad areas of solid color — if there's a way to do it without also making the fatty images that don't show artifacts even fatter, sounds like goodness.
Woot! I would be very interested in what you guys come up with. You guys are the best.
the middle shot, to me, seems a bit oversharp - as i've matured as a photographer i feel that not everything needs to cut your skin sharp! some stuff, yeah. i actually prefer the look of the third version, and to unsavory: there are no details lost - this is the original file as i processed it from the 16mpx canon 1Ds Mark II... anyhow, the sharpening thing is such a personal preference that we shouldn't debate that.
my point, that i'm attempting to make, is that i feel that the compression is acceptable for the -L images that i show in my galleries. could it be less compression? sure - and the quality would even go up - but we cannot ignore the tradeoff of file size vs. viewer experience (speed).
I have a lot to say here, so here goes. Firstly, when I said lost details I was referring to the Smugmug compressed one. Not the original file you processed. That one is very clear and sharp just as I would expect. It's the Smugmug one that looks soft and out of focus to me. As you've already noted, the sharpness of the photo is a personal preference. And this is precisely the point I am trying to make.
As an artist, I have a specific look I am trying to achieve in my photographs. I shoot in RAW, post-process, color correct, and get it just the right amount of sharpness. When I'm finished, I am happy with my work and ready to show the world. So I upload my photos and Smugmug compresses the images to the point where they are no longer as sharp as I want them.
Now to me, it is completely irrelevant that 90% of people cannot tell the difference. The important thing to me is having my work displayed EXACTLY as I intended it to be. I sometimes show my work to other photographers who CAN tell the difference. And it is a poor reflection of me and my work to have JPEG artifacts in it because it makes my photos look subpar. Not up to snuff. The difference between a clean crisp photo and the Smugmug compressed one is night and day to me.
Take a 5 star chef as an example. She uses the absolute best ingredients, spends hours and hours perfecting her recipes and nit-picks every last detail. Do you think that chef cares that 99% of the people that eat in that restaurant can't tell the difference between a truffle and a mushroom? No. Why? Because she wants to be the best chef in the world. And using subpar ingredients is not going to get her there.
This example is a little extreme because I am not trying to be the best photographer in the world. However, as a photographer I am continually trying to be the best I can be. And it's the small things that make a difference.
Now I know that 98% of Smugmug's customers don't know how to sharpen, compress or size their own photos, and I know that they can't tell the difference anyway. Most people use Smugmug as a snapshot repository. I myself did this for years. So I understand if they make the decision not to cater to the 1% - 5% that CAN actually tell the difference.
But if they are going to attempt to cater to Professionals and offer Pro accounts, they should allow their Pro users to have their photographs displayed EXACTLY as they would like them to be.
All the arguments I have seen in favor of Smugmug on this issue are making excuses for degrading the quality of an artist's work. "The medium image is too small to be a fair comparison." "They allow the ability to view the original." "You can downsize to below 800px and then it won't be compressed." Nevermind the fact that as an artist I don't want my work being displayed at ANY resolution that has horrible compression. It doesn't matter to me that the original is available. I don't want people viewing my work at medium size at all if it isn't going to meet my standards.
Is this so unreasonable? Is it such a big thing to ask that as a pro user I should be able to display my photos the way I want them? I really don't think it is unreasonable.
Or should I as a customer of Smugmug accept that they cannot supply me with a quality photo that meets my standards?
Please don't take all this the wrong way. I love all you guys/gals at Smugmug and think you are doing an excellent job at what you do. I'm just not convinced that you are capable of catering to the customers that have an expectation of having their work displayed exactly as they intended it. Not because you aren't technically capable (because you are), but because of business demands. Perhaps there is a conflict of interests in play here.
Comments
But I think that by-and-large, simply upping the qualiy of every single S, M & L image on one's website would greatly increase lag-time. All you DSL etc. users have NO idea how slow these images are loading for the 56k-ers out there. My girlfriend just recently visited her granparents, and she wanted to show them my photos but they basically had to stop after a while because I just have so many photos that it would have taken literally all day and all night. Now with my DSL connection I can show someone practially my whole portfolio / exploits galleries in 30-45 minutes. I know more and more people are converting to faster internet connections, but I think it's wise that smugmug is not just leaving the 56k-ers in the dust. If a McAskill (sp?) wants to chime in and give us some stats on the ratio of cable-to-dialup users on Smugug, I bet we'd be surprised.
I never thought I'd find myself saying this, but other than maybe, maybe, MAYBE adding a single image "refresh thumbnail" button, we definitely must leave image compression quality the way it is right now.
If you REALLY want higher quality images, then use a Photoshop action to take your high-res photos and make them 700-800 pixels, run your best shot of USM, and upload those photos into your original gallery. Disable printing for this gallery. Move the high-res photos to a new, "all thumbs" gallery, disable larges and title the gallery "such-and-such event/genre PRINTS gallery" and post a link to it in the original gallery. In my opinion that is a very professional means of displaying your photos, and if you want, it wouldn't hurt to explain to people why you've set things up this way. I bet you'd get lots of comments from people saying that your 800 pixel "high quality thumbnails" load very slowly!
Take care,
-Matt-
My SmugMug Portfolio • My Astro-Landscape Photo Blog • Dgrin Weddings Forum
Choice, choice, choice. Customers like choices. Something Smugmug does a terrific job at, with exception to this issue.
While I can see the artifacts in my resized images, I don't particularly mind. But I don't think that my not minding is a good reason to tell you not to mind. If indeed everyone is as happy with their jpeg compression as they say they are, then adding the capability to allow certain pros to increase it could possibly even be amortized across the entire range of smugmug customers and the change would be nearly imperceptible (a bit like how the "infinite" storage gets amortized out.) A pro acknowledges that decreasing compression increases size (who doesn't?) and would take longer to download. But they are grown adults and know who their audience is, and will generally make an informed decision about load times if they value image quality enough to want to compromise them.
I don't know how smugmug maintains its huge database of photos, but I can't think of any scheme by which this would be a difficult change, technically. The argument against it, that I've heard, is that people will go complaining to smugmug if the load times are too slow. I wouldn't think so, since most people will think they're visiting the photographer's site, not the photographer's site hosted on smugmug. They will tell whomever they think is the owner of the site (in this case the pro photographer) that they think their site is too slow. If the pro gets too many of these emails, he/she might consider increasing the compression again. But allowing this change doesn't seem like such a big deal.
All this arguing against the change seems, in some cases, inspired by cognitive dissonance... but I may just be reading things into it that I shouldn't
Just my 0.02 eurocents
Dan
ok -
Portfolio • Workshops • Facebook • Twitter
But that also means that people can take those images and print them themselves. The other option is to upload a <800px image and then you necessarily lose many of the benefits of a pro account (mostly being able to sell photos online for a profit.) Or the third option is to just accept the artifacts, which doesn't seem like a very good one either.
Isn't it like buying a new car and finding a scratch on it? You know it won't affect the car's performance in the least, but it still bugs you if you know about it and see it every time you use the car.
The whole point of wanting crisp photos is so that people browsing my galleries will see my photos the way I meant to have them seen. I didn't spend hours post-processing my photos, getting them just the right sharpness, sizing to 1024, etc only to have them viewed all jagged and fugly. I really don't think this is too much to ask. Like I said, I would be willing to pay a premium for this.
By the way, I completely agree with what pumpkin has said about the affect of allowing this option to certain pro users being amortized across their entire user base.
What they should be doing if they are interested in good image quality and fast downloads is targeting a constant thumbnail file *size*. This is what pretty much every digital camera on the market does. Take a picture of static on your TV or a clear blue sky and the file sizes come our remarkably similar in size. This of course seems impossible since there are vastly different amounts of information in each image, until you look at the JPEG headers and you'll discover the quatization matrix for each image is drastically different - the blue sky less lossy (less rounding in the quantization matrix) and the static more lossy (more rounding).
So the general idea is to adjust the quantization matix (quality setting) until you get the desired size. This results in much nicer looking photos. Why? Well, in very complicated images that require lower quality settings to meet the target file size the artifacts are less visible - the complicated scene information hides them. In images with large uniform patches at the lower quality settings artifacts would show up like mad at the edges (sky for example) - but large uniform patches compress real well so the file can have a much higher quailty setting and still meet the target file size.
So far most of the example images posted in threads like these by people who feel their photos were mauled by the compression have had large uniform areas in them (sky) and the artifacts are all visible in the sky. And if you look at the file *sizes* they are way smaller than they need to be. Here's an example from one of the few non-private galleries I've got up at the moment:
http://www.kenandchristine.com/gallery/615329/1/29198141/Large
There are some scary artifacts around the lightning. This image has lots of uniform space and had noise reduction applied to it, as a result it compresses well. The large image is only 30KB!!!! Unfortunately from the clouds to the bolt is a big step function and artifacts show up real nice in that area. A higher quality setting would remove the artifacts and still keep the image at a managable size.
http://www.kenandchristine.com/gallery/615329/1/26081308/Large
Now this image is 70KB, more than *twice* the size of the first!!! It's contrasty and there are few uniform areas. You can pick up a little artifacting where the Sierra's meet the sky, but its not objectionable, and everywhere else in the image the artifacts are hidden by other scene information (e.g. the top of the dunes is a very sharp edge, but the artifacting is much harder to see than in the lightning image).
Long story short, the current constant quality settings are suboptimal for both image quality and download time. The quality should be set adaptively to achive a target file size for the thumbnails. This is a back end change for the smugmug folks (we love you, by the way, it's just tough love
Anywho, this has come up before, I think even Baldy mentioned they were looking into it. I sure hope they do. It will result in both good bandwidth performance and image quality. It can't be that hard to do if your camera is doing it at 5 fps. I think the problem is many of the command line processing tools out there don't have a target file size option so they might need to wrap a script around whatever they are using to do it right, and as we know it isn't like they're sitting on their duffs drinking beer at the moment.
Oh, and I think I mentioned this on another thread, if you are truely disturbed by what are in fact pretty minor artifacts I strongly suggest you don't visit your clients or family and friends who are viewing your works of art on crappy uncalibrated monitors. The drastic color shifts are *way* more objectionable.
Ken
the poster said his constituency was friends and family.
Portfolio • Workshops • Facebook • Twitter
i wonder how many others would? i'll make sure this gets consideration ... thanks so much for your input and suggestions!
Portfolio • Workshops • Facebook • Twitter
an excellent suggestion!
this discussion is fascinating...really... i'm looking forward to your answers
three versions of the same shot. all from a canon 1Ds Mark II shot, the full-size jpg is just under 8 megabytes.
which one was made by smugmug, and which was made by me, and finally, which one was made by me and also compressed by half?
(no cheating by looking at the file size!)
Portfolio • Workshops • Facebook • Twitter
It's the pixelation around the necklace/sky and shirt/water border that gives it away.
http://photos.mikelanestudios.com/
why are you showing -M? i mean, c'mon - with most any camera today, you're talking a huge amount of difference between a -M size and the original
Portfolio • Workshops • Facebook • Twitter
http://photos.mikelanestudios.com/
ahh now you get to a whole 'nother can of tomatoes: sharpening! more later
Portfolio • Workshops • Facebook • Twitter
unsavory, what's your workflow? i'd also like to see a shot that you process, full size, and upload to smugmug and then link the -L and also a home-brewed 800px version. thanks!
Portfolio • Workshops • Facebook • Twitter
Am I just not being ciritical enough?
http://photos.mikelanestudios.com/
a big part of this whole issue is that photographers will always notice this stuff more than buyers
Portfolio • Workshops • Facebook • Twitter
http://photos.mikelanestudios.com/
so, i've put the answers here. mike you were spot on - and so here are my comments.
the middle shot, to me, seems a bit oversharp - as i've matured as a photographer i feel that not everything needs to cut your skin sharp! some stuff, yeah. i actually prefer the look of the third version, and to unsavory: there are no details lost - this is the original file as i processed it from the 16mpx canon 1Ds Mark II... anyhow, the sharpening thing is such a personal preference that we shouldn't debate that.
my point, that i'm attempting to make, is that i feel that the compression is acceptable for the -L images that i show in my galleries. could it be less compression? sure - and the quality would even go up - but we cannot ignore the tradeoff of file size vs. viewer experience (speed).
fascinating discussion!
Portfolio • Workshops • Facebook • Twitter
One thing that confuses me is I'm looking at a folder of differnet shots from my 20D, all at the same medium compression setting, and they range from 606k to 1518k depending on the scene I understood kwalsh to say virtually every digital camera puts out a constant size image (he said thumbnail, but we're talking more than thumbs, correct?).
ken, a great post with some excellent food for thought. and your final point is soooooo true - i have seen some folks with multi-thou$and dollar displays and they look like the dog barfed on the screen it's so green. :puke
Portfolio • Workshops • Facebook • Twitter
Beware small number statistics for they shall make you look like a moron! :crazy
So, yeah Baldy you're right on that one. "Virtually every" was strong language for me to use on a survey of 5 cameras. Turns out none of them were from Canon. Canon uses a constant quantization matrix. So, for a larger survey across more manufacturers and cameras:
Canon 10D - Constant
Canon 2is - Constant
Canon G3 - Constant
Canon S70 - Constant
Sony DSC-H1 - Dynamic
Sony DSC-s85 - Dynamic
Sony DSC-L1 - Dynamic
Sony DSC-707 - Dynamic
Nikon 8800 - Dynamic
Nikon 4500 - Dynamic
Nikon D100 - Dynamic
Fuji S3 Pro - Dynamic
Fuji F700 - Dynamic
Minolta Z5 - Constant
Minolta 7Hi - Constant
Olympus C-8080WZ - Dynamic
Olympus E-10 - Dynamic
So, it would appear that for any given manufacturer they've either set compression dynamically or constant across their product line more or less since the dawn of time (there are some old cameras in that list). Dynamic seems to be the more common, but by no means "virtually every". Of course if you take your five samples from Nikon, Sony and Olympus it'll look that way
What do you guys use to convert on the back end (if you're comfortable saying)? If it's freeware I'd love to try out some simple scripts to see how easy/hard a target file size scheme is on my own system.
Ken
P.S. Hope your headache is better.
We use Image Magick. I talked to Onethumb last night and I think we're pretty open to making the small images (byte-wise) bigger — the ones with the broad areas of solid color — if there's a way to do it without also making the fatty images that don't show artifacts even fatter, sounds like goodness.
Ken
Woot! I would be very interested in what you guys come up with. You guys are the best.
I have a lot to say here, so here goes. Firstly, when I said lost details I was referring to the Smugmug compressed one. Not the original file you processed. That one is very clear and sharp just as I would expect. It's the Smugmug one that looks soft and out of focus to me. As you've already noted, the sharpness of the photo is a personal preference. And this is precisely the point I am trying to make.
As an artist, I have a specific look I am trying to achieve in my photographs. I shoot in RAW, post-process, color correct, and get it just the right amount of sharpness. When I'm finished, I am happy with my work and ready to show the world. So I upload my photos and Smugmug compresses the images to the point where they are no longer as sharp as I want them.
Now to me, it is completely irrelevant that 90% of people cannot tell the difference. The important thing to me is having my work displayed EXACTLY as I intended it to be. I sometimes show my work to other photographers who CAN tell the difference. And it is a poor reflection of me and my work to have JPEG artifacts in it because it makes my photos look subpar. Not up to snuff. The difference between a clean crisp photo and the Smugmug compressed one is night and day to me.
Take a 5 star chef as an example. She uses the absolute best ingredients, spends hours and hours perfecting her recipes and nit-picks every last detail. Do you think that chef cares that 99% of the people that eat in that restaurant can't tell the difference between a truffle and a mushroom? No. Why? Because she wants to be the best chef in the world. And using subpar ingredients is not going to get her there.
This example is a little extreme because I am not trying to be the best photographer in the world. However, as a photographer I am continually trying to be the best I can be. And it's the small things that make a difference.
Now I know that 98% of Smugmug's customers don't know how to sharpen, compress or size their own photos, and I know that they can't tell the difference anyway. Most people use Smugmug as a snapshot repository. I myself did this for years. So I understand if they make the decision not to cater to the 1% - 5% that CAN actually tell the difference.
But if they are going to attempt to cater to Professionals and offer Pro accounts, they should allow their Pro users to have their photographs displayed EXACTLY as they would like them to be.
All the arguments I have seen in favor of Smugmug on this issue are making excuses for degrading the quality of an artist's work. "The medium image is too small to be a fair comparison." "They allow the ability to view the original." "You can downsize to below 800px and then it won't be compressed." Nevermind the fact that as an artist I don't want my work being displayed at ANY resolution that has horrible compression. It doesn't matter to me that the original is available. I don't want people viewing my work at medium size at all if it isn't going to meet my standards.
Is this so unreasonable? Is it such a big thing to ask that as a pro user I should be able to display my photos the way I want them? I really don't think it is unreasonable.
Or should I as a customer of Smugmug accept that they cannot supply me with a quality photo that meets my standards?
Please don't take all this the wrong way. I love all you guys/gals at Smugmug and think you are doing an excellent job at what you do. I'm just not convinced that you are capable of catering to the customers that have an expectation of having their work displayed exactly as they intended it. Not because you aren't technically capable (because you are), but because of business demands. Perhaps there is a conflict of interests in play here.
Thoughts?