PDA

View Full Version : Achieving great JPEG Compression


Godly
11-06-2006, 12:16 PM
I really would like to hear some detailed explanations and advice/tips from you guys on how to improve the final product of a photo when limiting yourself to specific filesizes. I don't know where the differences mostly come and what people can do to make the most of those elements to get a better photo in the end with a lower filesize. How much impact does the initial resolution (after a certain point in camera mp, of course), file format that's post-processed, and the processing/conversion software used have on final photo's compression and quality?

For example, I would save every photo at under 100k. Are there things that could be done to make it possible to save a photo to be under 100k and at a higher quality JPEG depending on the software you're using and/or the other things I asked above? I just wing my own process since I don't really know what I'm doing, but here's my current process:
1) Canon 20D RAW format
2) Photoshop CS - RAW Conversion to uncompressed TIFF
3) Photoshop CS - Edit TIFF
4) Photoshop CS - Save for web and make filesize as close to 100k as possible and save JPEG

Now, many a time I have to save the jpg at say a quality of 50-60 if there is a lot of detail/colors in the photo and of course that JPEG looks really bad to me. I'm wondering if that's pretty much how it goes or is there a way to get that same photo to come out at a higher quality JPEG with the same filesize by changing the software used or hardware even? Like, does higher MP actually enable a better JPEG compression in the end in this instance (going from like 8mp to 10 or 17)? Does editing in RAW instead of TIFF and then saving directly to JPEG make a difference? Does the software used for editing and/or conversion of RAWs make a difference?

Boggles my mind.

Thanks in advance. I'm sure I'll get good information from you guys.

Spetsnaz
11-06-2006, 02:05 PM
Using photoshop is a start. I usually select jpeg and scroll the quality bar to what I want. Make sure you save two copies of your picture. One of the original, and any you want to share online as the JPEG.

You can do some editing on the pictures and make them look REALLY good by changing lighting in them and keeping the file size low.

http://img48.imageshack.us/img48/5240/zoe2xz7.jpg

The original for this picture was over 300 KB. The picture was also taken with a 2.0 Megapixel camera (Believe it or not.) While I have a much better camera now, which takes pictures in much higher KB range (500-600KB at times) I can still achieve a small file size and make it look good.

What I did to that picture was go into photoshop...went into image, adjustments, and toyed with the 'curves' option. Then I saved it as a JPEG on medium settings. If there was any sort of image quality reduction, my editing hid it well by playing with the lighting.

Also, if you pictures are coming in high resolutions...simply reducing the pictures to 800x600 Pixels will drastically reduce the filesize...and 800x600 is still a good size to show online. Of course, there is also 1024x768 which will also be good for showing off pictures and will reduce your file sizes.

Tinker with resolutions, lighting and with the 'quality' bar found when you save on photoshop and you should be just fine.

Godly
11-06-2006, 02:15 PM
Hey Spetsnaz, I appreciate the input. However, my actual question is a bit different than how you interpreted it and with the situation you're working with, I think you'll find the answer just as useful for you as it will be for me.

I see staereo is in this thread so perhaps he's working on a long, detailed post with some answers and perhaps example comparisons.

staereo
11-06-2006, 02:34 PM
I really would like to hear some detailed explanations and advice/tips from you guys on how to improve the final product of a photo when limiting yourself to specific filesizes.

I use the same method as you for the most part.

However, I run raw conversion output based on my final products required. Hi quality digital will be an 8/16 bit TIF in sRGB space, print will be 8/16 bit TIF in adobeRGB space. Low res proofs or for cosplay galleries will be directly saved as a sRGB JPG, low compression. If I am running more than one final product, then I end up running through post processing more than once on the same image. But I don't use TIF output for cosplay galleries, really. Just too much wasted space, when much of the benefits of a lossless TIF file offers ends up lost when its saved for web as a small sized, highly compressed jpg anyways. So this is ends up being a convenience thing more than anything.

I don't know where the differences mostly come and what people can do to make the most of those elements to get a better photo in the end with a lower filesize. How much impact does the initial resolution (after a certain point in camera mp, of course), file format that's post-processed, and the processing/conversion software used have on final photo's compression and quality?

File size has more to do with just how many pixels are recorded. The dynamic range of a camera's processing has a ton to do with file size, as the shades, range, depth of color that a processor has greatly increases the file size on its way out. A camera that records 3 stops of range in an image will inherently use less space than a camera that records 5 stops of range, when there is 5 stops of range in an image. Typically, more data is stored in lighter areas in an image, as there are more levels of data in light than there are in dark. This is because unlike emulsions, image sensors record in a linear fashion. This means if your lightest stop has 2048 levels, then the next stop darker will have 1024 levels, your next stop darker will have 512 levels, your next stop down will have 256, and the darkest stop will have 128 levels. Because you often aim to have your middle stop be your midtone, your end up putting a lot of your memory into lighter levels. For this reason, many people choose to 'expose to the right', and that means that they record a near-overexposed image with the understanding that they can retain more levels of data using the lighter stops of their exposure, so long as they don't lose detail in the overexposure. Anyways, long story short is: the more stops of light your range is (be it camera limited or light/color limited) the larger your file. Megapixels are distinct file-sizers, but when considering your resizing down, the actual resolution of the picture is not an issue at capture. It is more to do with the dynamic range at capture, and whether or not that dynamic range is carried on to your web-targeted image.

If you're capturing RAW, then the following doesn't really apply, but to see an idea of color range in action, see if your camera can record jpeg in aRGB and sRGB. Take a picture of a green rich picture in both color spaces, and see which is a larger file. This shows how the range of color can affect your image size. Again, this will only change if your software removes color range in the conversion to a web targeted image. So thats another factor.

The program by which you convert to a web-targeted image will have a lot to do with your output. As you can see, depending on what you choose to use, you can strip a lot of data. Photoshops ImageReady 'save for web' is pretty smart. It strips what you don't need: exif/metadata/etc, converts to sRGB (I think??), and does a good job at remaining accurate to the original. By accurate, I mean it attempts to keep the colors and such as close as it can without doing much 'guessing'. Sometimes this leads to a less 'punchy' output, sometimes images look a little faded and such. Other software tends to 'guesstimate' on what the image SHOULD look like, and it boosts levels and such to compensate for stripped data. I prefer to tune those items myself after the resizing, as I am picky about data in my image. Either way, you can see how it all depends on your software. As I said, photoshop is pretty well 'fantastic' at downsizing images as accurately as possible.

but here's my current process:
1) Canon 20D RAW format
2) Photoshop CS - RAW Conversion to uncompressed TIFF
3) Photoshop CS - Edit TIFF
4) Photoshop CS - Save for web and make filesize as close to 100k as possible and save JPEG

Good process. Nothing wrong with that. If this is what you're doing, theres nothing wrong with that. You may find that a resize prior to save for web is better, you may not. Either way, this process is perfectly legitimate. Perhaps keeping an eye on your colorspace and color depth in TIF, and using an sRGB, 8bit colorspace through your entire process will lead to more pleasing results at the end. (Though chances are, you're already doing this.)

Now, many a time I have to save the jpg at say a quality of 50-60 if there is a lot of detail/colors in the photo and of course that JPEG looks really bad to me. I'm wondering if that's pretty much how it goes or is there a way to get that same photo to come out at a higher quality JPEG with the same filesize by changing the software used or hardware even?

You can try other software titles, but they end up sacrificing accuracy for guesstimated assumptions at what an image should look like. You can see if other titles offer guesstimations that are ok with you, but for the most part to keep on par with good accuracy, you're about right on with saying that's pretty much how it goes.

Like, does higher MP actually enable a better JPEG compression in the end in this instance (going from like 8mp to 10 or 17)?

No. Answered more in depth above. Either way, when shooting raw you can convert this depth to anything you choose before hitting an image file format.

Does editing in RAW instead of TIFF and then saving directly to JPEG make a difference?

Not really. If you use jpeg and save time and time again, you will see a decrease in image quality for each time you save it, as it re compresses each time. Think copy of a copy. But this more or less affects the quality that you end up getting at the same 'level of compression'. RAW => TIF => Jpeg when talking a one time save-for-web deal will not really get you any better results than any other process with the same final product. TIF is lossless compression.

Does the software used for editing and/or conversion of RAWs make a difference?

As stated above, yes. But quality and accuracy play huge roles. Raw conversion software all convert image and 'see' RAW files as different images, so your output is different in every single one. So will your file sizes. This ends up being a personal preferance deal more than anything else. PP software like photoshop, play a HUGE role in size of your image files when saving for use on the internet. I explained this in more detail above. Personal opinion for me on this is let my images suffer slightly on quality, and make up for the quality with accuracy.

Boggles my mind.

Mine too... Mine too.

I hope some of this was informative. Feel free to let me know if I can explain anything further, and if I can I will try.

Bruce

staereo
11-06-2006, 02:39 PM
I see staereo is in this thread so perhaps he's working on a long, detailed post with some answers and perhaps example comparisons.

I about died laughing when I read this. Touche. I am far too predictable. I need to start tossing out some one-word replies here. :bigtu:

On that note, I can go ahead and do some examples and post them up here when I get home.

I am very much a visual and hands-on learner, so I might be able to show examples of software types this way, if you'd like?

Bruce

Godly
11-06-2006, 02:59 PM
Thanks for the detailed info. I'll reread it over and over (especially when I get home) so I know I fully understand it (or find out I don't fully understand it).


Good process. Nothing wrong with that. If this is what you're doing, theres nothing wrong with that. You may find that a resize prior to save for web is better, you may not. Either way, this process is perfectly legitimate. Perhaps keeping an eye on your colorspace and color depth in TIF, and using an sRGB, 8bit colorspace through your entire process will lead to more pleasing results at the end. (Though chances are, you're already doing this.)

Yes, I actually am doing these things. I didn't go into the details of postprocessing as I felt the order in which I do my editing isn't the biggest cause of the problem. I also think I'm doing it in one of the best orders already anyways. Since it seems like it'll help in this thread though, I'll go into it:
1) RAW -> TIFF *I actually remove all or most autoadjustments that PS does on the RAW and do the editing as fresh as possible on the TIFF
2) Editing TIFF:
- Noise reduction if necessary
- Shadows/Highlights (Curves)
- Color Balance
- Any touching up (Dodge/Burn/Sponge, Cloning, etc)
3) Save TIFF
4) Resize to web res (640px)
5) Sharpen (USM)
6) Watermark
7) Flatten (watermark)
8) Save for Web to under 100k

Yea I do know about one time saving to JPEG and I always practice that. Any additional changes go back to the TIFF to recreate the JPEG.

An example of a photo that gets saved at quality 50's in JPEG in my process to maintain an under 100k filesize would be like this:
http://img.photobucket.com/albums/v605/godly/Trinity%20Blood%20-%20Private%20Shoot/pvtshoot-trinityblood-26.jpg

I'm trying to understand what steps could've been taken to increase the quality for that filesize for this photo.


On that note, I can go ahead and do some examples and post them up here when I get home.

Yes please!

I think I had something else to say, but I'm going to lunch so I'll continue when I get back.

Thanks much!

staereo
11-06-2006, 03:34 PM
You have a very well thought out process. You do everything properly. I don't typically (if ever?) use acr, so I'm not entirely sure what it's abilities are. This isn't because I am opposed to acr, it just happens that other software fits my personal workflow better. exposure, saturation, contrast, and WB are often good to adjust while in raw.

Sometimes I raw presharpen, but this usually occurs when im hitting the direct jpeg workflow for smaller pics, as detail really isn't as important, and P1C1 does a pretty good job of presharpen (as does nik sharpener 2 in acr, again, haven't really fooled with that enough to comment.).

Sharpening last in your typical workflow is spot on, and is often something I leave UNSAVED in my images. Reason being is I use nik, and I sharpen for target print/display. So if I use a lab, home printer, or send for press they all receive different sharpening. The couple of minutes it takes to sharpen an image up isn't really worth a save each time, so I just leave images unsharpened until they reach their final destination. This appears to be your workflow as well, so we are on the same page.

I'm not near any of my software at the moment, so pardon me using my eyes to toss ideas out. First of all, you have a huge dynamic range in that image. Check out the darkest details in the image, then the lightest details. You're using all of the stops of range that your camera has to offer. I can spot both overexposure and underexposure in the image. This means BIG file. (Not something you should necessarily change, though.)

Next, take a look at how much data lays in the lighter portion of your image. WOW, so you're really taking advantage of those layer-rich light levels of your camera's range. Another file space inflator. Theres a lot more data in those light details of the image, as it is a linear recording of light. (Again, not really something you should be changing unless you're changing the look of the picture.)

Next, look at the color RANGE you offer. Red, green, blue.

You could just look right at your histogram and take a guess that this file is going to eat space, this is just a large file with a lot of data. It is no wonder you need to knock the quality down so low to fit it under 100k.

I have found that personally, dropping my long-side to 600px and increasing my personal file size limitation to 120k really gives me more breathing room for when I come across file-monsters like this. The loss of compression makes up immensely for the lack of those 40 px on the long side. Just a personal observation though, you may find a different 'sweet spot' in your own research. When cropping images away from the 2:3 image ratio, so they become more square, I tend to drop the longer side further. maybe down to 500-550px, just to drop my need for compression. A little loss in size of your px goes a long way for keeping a full, rich, vibrant and smooth output.

In addition, try maybe a little less sharpening when you KNOW you're going to have to go heavy on your compression. When you get to checking your dynamic range in an image, or your color range, you will get an idea beforehand that your image is going to be a size-hog. When you know this, try dropping your sharpening a little. While your image MAY be a little softer (let's face it, at that size a blurry image will get sharp), the jpeg artifacts will appear lessened because they will not form as strongly around softer lines and color gradients. This will at least take one piece of the 'ickyness' of the compression out of the puzzle. But play around for yourself. Output is really a subjective thing. It is purely my opinion that jpeg compression artifacts make a good sharpen-job look oversharpened. Others can easily see it differently, you know?

Anywho, I'll be around for a bit. I'm always willing to shoot the perverbial... well, 'stuff' when it comes to photography.

I'll also check into tossing up some images on my server and posting them here with a few different size examples.

Bruce

Godly
11-06-2006, 04:35 PM
Hey, thanks much for all the helpful detail. I guess from the pointers you gave, the concensus is that the situation itself is normal and there is no magical tool that I was wondering about (hoping for?). I understand all the possible reduction methods you gave and yea, I pretty much did what I did with that file because that is how I actually want it (in terms of going against a softer photo and the colors/contrast used).

Having never used high end hardware nor high end converters I was curious as to if it would change this specific issue I had, and now I know what I considered an 'issue' isn't really an issue. It's just common sense about how small 100k is and how large 640px is and how much of a b*tch JPEGs are =).

I think I was just under the impression that it was possible because I thought I would see photos from like Admin and think "That seems to have just as much 'stuff' that would bump filesize up as a highly compressed photo I took has, yet it looks like a lot more was retained in that small filesize and same imagesize." Perhaps I wasn't looking closely enough and/or was looking at the wrong factors.

So, I guess I still am interested in seeing comparison photos of the same processes done in different converters and/or edits done on RAW vs TIFF, but if there isn't a tremendous difference then I can ignore that particular itch.

Questions from this:
What' acr? What's P1C1? What's Nik Sharpener?

Godly
11-06-2006, 05:04 PM
Oh, also can you tell me the difference between Imageready Save for Web and PS Save for Web? In fact, I never use Imageready as I don't know its application. Sorry for asking you instead of looking it up.

staereo
11-06-2006, 08:42 PM
So, I guess I still am interested in seeing comparison photos of the same processes done in different converters and/or edits done on RAW vs TIFF, but if there isn't a tremendous difference then I can ignore that particular itch.

Questions from this:
What' acr? What's P1C1? What's Nik Sharpener?

Ill go ahead and show you some comparisons, I just got to sitting down at my pc, so bear with me. I'll use some shots of waterfowl I took a couple of days ago, since I haven't looked at them yet.

ACR = Adobe Camera Raw (If you convert in adobe cs, this is what you convert with.)

P1C1 = Phase One Capture One. Raw conversion software. I use the pro version.

Nik Sharpener = A high end photoshop plugin. Runs 200 bucks for the plugin, and allows you very finely tuned sharpening tools that can use colors, as well as output based on final print size, viewing distance, and printer profile type. In addition, allows for pre sharpening and web sharpening. Very intuitive. Somewhat overkill in the sharpening world, but has its place.

CS's save for web is part of a CS program group called Image Ready. So, the one you use in CS is what I'm referring to. Sorry for confusion.

Ill get to those images now, will post in a bit.

Bruce

staereo
11-06-2006, 10:55 PM
Well, this took a couple of hours. I'm tired.

I used 2 raw converters. Sorry, but I didn't use Adobe.

I used canon digital photo pro, and phase one capture one.

I then took images from each, and attempted to create nice images out of both. I converted them to 8 bit TIFs in sRGB colorspace. I took the images and loaded them into cs2 to create similar crops in each.

From there, I ran them through conversion software that manages jpeg compression. I resized and cropped them in the same piece of software.

I used Photoshop's Image Ready, PhotoWatermarkPro, and ACDSee to resize the images and compress them into jpegs. I tried to keep size at 600 longest side, and kept file size at 100k and under. (some went slightly over in acdsee as it was a guess and check on file size, and Im tired.)

CaptureOne

Photoshop's ImageReady
http://www.staereo.com/cap1/ir/_MG_0030.jpg
http://www.staereo.com/cap1/ir/_MG_0034.jpg
http://www.staereo.com/cap1/ir/_MG_0175.jpg
http://www.staereo.com/cap1/ir/_MG_0177.jpg
http://www.staereo.com/cap1/ir/_MG_0187.jpg
http://www.staereo.com/cap1/ir/_MG_0195.jpg

PhotoWatermarkPro
http://www.staereo.com/cap1/pwp/_MG_0030.jpg
http://www.staereo.com/cap1/pwp/_MG_0034.jpg
http://www.staereo.com/cap1/pwp/_MG_0175.jpg
http://www.staereo.com/cap1/pwp/_MG_0177.jpg
http://www.staereo.com/cap1/pwp/_MG_0187.jpg
http://www.staereo.com/cap1/pwp/_MG_0195.jpg

ACDSee
http://www.staereo.com/cap1/acd/_MG_0030.jpg
http://www.staereo.com/cap1/acd/_MG_0034.jpg
http://www.staereo.com/cap1/acd/_MG_0175.jpg
http://www.staereo.com/cap1/acd/_MG_0177.jpg
http://www.staereo.com/cap1/acd/_MG_0187.jpg
http://www.staereo.com/cap1/acd/_MG_0195.jpg


Digital Photo Pro

Photoshop's ImageReady
http://www.staereo.com/digpp/ir/_MG_0030.jpg
http://www.staereo.com/digpp/ir/_MG_0034.jpg
http://www.staereo.com/digpp/ir/_MG_0175.jpg
http://www.staereo.com/digpp/ir/_MG_0177.jpg
http://www.staereo.com/digpp/ir/_MG_0187.jpg
http://www.staereo.com/digpp/ir/_MG_0195.jpg

PhotoWatermarkPro
http://www.staereo.com/digpp/pwp/_MG_0030.jpg
http://www.staereo.com/digpp/pwp/_MG_0034.jpg
http://www.staereo.com/digpp/pwp/_MG_0175.jpg
http://www.staereo.com/digpp/pwp/_MG_0177.jpg
http://www.staereo.com/digpp/pwp/_MG_0187.jpg
http://www.staereo.com/digpp/pwp/_MG_0195.jpg

ACDSee
http://www.staereo.com/digpp/acd/_MG_0030.jpg
http://www.staereo.com/digpp/acd/_MG_0034.jpg
http://www.staereo.com/digpp/acd/_MG_0175.jpg
http://www.staereo.com/digpp/acd/_MG_0177.jpg
http://www.staereo.com/digpp/acd/_MG_0187.jpg
http://www.staereo.com/digpp/acd/_MG_0195.jpg

Gosh I hope this helps. Im too tired to even see the results. Good luck!

Bruce

Ollie
11-06-2006, 11:55 PM
The best way to achieve better image quality with the same file size is to use a better compression algorithm. JPEG2000 is a standand that hasn't really caught on yet. However, JPEG2000 uses a wavelet transform instead of the 8x8 blocks JPEG uses. It boasts 10 more compression for the same image quality. It's really only like maybe 2 or three times as much, but still that's a big improvement. I view this from the purely theoretical side of things, so this probably isn't too useful, but the whole digital aspect of photography wouldn't be possible without a lot of very facinating math behind it. If you read up on it, you can learn a lot about what the mind actually sees as opposed to what is in a picture.

shiroin
11-07-2006, 01:05 AM
for rates of compression
i find 91% or around 10 our of 12 (on the photoshop scale) to be the most optimal interms of quality/file size ratio.
anything above that does not achieve significant visible quality but dramatically increases file size.

for algorithms use baseline optimized if it is an option. this will give you a 10% decrease in file size without much lost in detail.

if you desire no quality loss i suggest TIFF with LZW compression. (ZIP compression is not supported by many softwares and brings roughly the same file size)

Godly
11-07-2006, 01:07 AM
Thanks so much for all of that work staereo! Currently, I have a hard time finding out which photos are better than the other to me. Those all seem to have been saved at pretty high jpeg qualities though so I don't know if that just means a 50 quality jpeg would be similar (as in, they'd all be around 50 and look the same) or a completely different situation. I'll study your comparisons a lot more. Thanks again.

Ollie, any details you have onhand to share? =)

jtnishi
11-07-2006, 01:08 AM
The best way to achieve better image quality with the same file size is to use a better compression algorithm. JPEG2000 is a standand that hasn't really caught on yet. However, JPEG2000 uses a wavelet transform instead of the 8x8 blocks JPEG uses. It boasts 10 more compression for the same image quality. It's really only like maybe 2 or three times as much, but still that's a big improvement. I view this from the purely theoretical side of things, so this probably isn't too useful, but the whole digital aspect of photography wouldn't be possible without a lot of very facinating math behind it. If you read up on it, you can learn a lot about what the mind actually sees as opposed to what is in a picture.
JPEG2000 is mostly compatible with web browsers, however I'm pretty certain that from a full compatibility perspective, it's probably better to stick with good ol' fashion JPEG.

As far as reducing size, I'm sure you already know the usual stuff, Godly. Noise reduction definitely helps (reduces noise detail that adds to size). Not sharpening as aggressively helps (or conversely, intentional softening reduces file size). Anything that works against areas with lots of details helps. Keeping resolution down helps. Beyond that, it probably gets tiresome: you can define masks in Photoshop such that you can save certain parts of an image in higher quality, while highly compressing the rest. But since you shoot portraiture style, that isn't likely to help much at all, not to mention it's way too time consuming.

staereo
11-07-2006, 06:00 AM
Those were saved at between 70 and 94 quality in photoshop's save for web.

Other programs were similar amongst their scales. When applications offered metadata stripping, I chose to do so.

I'll go ahead and help you out with noticing differences in compression.

image 0030.jpg had very little compression. I was able to save in the top level of quality almost every time I compressed it, desptie the program. This means it was hard to spot differences. Why? Well, you can probably tell that the dynamic range is small. Theres no underexposed portion to the image. The other reason is it offers very little in terms of color range. So it's small to compress anyways.

Image 0034.jpg offers a little insight. I chose to use it because it had a broad range, but with very little color in between. Knowing that it would be a little more heavily compressed, I wanted to show a lot of hard lines stacked up in the compression. You'll notice the pwp copy of this photo has many halos behind the right bird. This causes an appearance of oversharpening and just general artifacts added to the image. ACDSee isn't nearly as bad, as it's main fault here is the shortening of shadows in the overexposed water, dropping the range of tonal value in the added contrast.

Image 0175.jpg lets you see some haloing under the right bird's wing in the pwp copy. Pwp's compression also causes that same artifacting problem beneath the middle bird. ACDSee is far better, but suffers from haloing where colors contrast. A place you can see this is above the right bird's head, where the green of the feathers meets the blue of the water. You'll notice the head's halo. IR's seems to lack these flaws.

Image 0177.jpg is fantastic for seeing the differences. Look at the duck in the foreground. PWP's copy, when compared to either other, has serious artifact problems in the green, and more loss of detail in the white feathers of the underbody. Not to mention it softens the heck out of color. ACDSee AND PWP suffer bad halo problems on the background duck's head.

Image 0187.jpg didn't help a whole lot, as there wasn't hardly any compression done on these images. It wasn't a big file, because it didnt have a lot of exposure range. Theres really nothing underexposed in the image, and the whole image is generally overexposed.

Image 0195.jpg you will notice a loss of detail in the fence fabric in the foreground when looking at pwp versus the other two. It also shows artifacts in pwp's fence fabric. In ACDSee's image, it is a lot better quality, but you get color artifacts when you look about half way up the image on the left, where the ceiling meets the left upper wall. IR's lack these issues.

I hope this helps point you to some of the differences in quality between the compressions. But, like I said, on average I was saving between mid seventies and mid nineties in quality. This is about on par with most images I save.

Bruce

shiroin
11-07-2006, 02:00 PM
Other programs were similar amongst their scales. When applications offered metadata stripping, I chose to do so.

I am personally against stripping of metadata. Supplying metadata gives interested viewers a brief sense of how this shot was achieved, and thus will benefit from it.

staereo
11-07-2006, 02:56 PM
We're discussing how to keep as much quality in an image as possible when compressing. So whether for or against stripping of metadata, if you're trying to save every bit of information for graphic preservation, stripping metadata is a valid method to retaining quality at minimal file sizes.

I don't tend to care much whether metadata is in my images or not. In reality, it is noones business how I got an image. Sometimes I leave it, sometimes I dont. Sometimes I digitally mark in the metadata, but it really isnt a concern for me in cosplay photography.

HOWEVER, no matter one's opinion on metadata, it is a stripable peace of file size to consider if you are purely concerned about maxamizing every bit of quality out of a compressed image at a given size.

Thats all I was meaning this time.

Bruce

Godly
11-07-2006, 03:12 PM
I appreciate all the info given in the thread. I was hoping there was perhaps software/hardware that helps for this topic, but now I know I have to change the things I am being stubborn about in a give and take manner. I am a bit OCD on numerical consistency so having photos sized differently or at different ratios bothers me greatly (especially in the same set), but downsizing would be a necessary sacrifice to gain higher jpeg quality.

Intentially softening is a likely option I'll start practicing as even if you don't, the jpeg compression can create pixelation anyways that softening could've prevented. My only gripe with it would be having to create a couple of different TIFFs as I do my noise reduction as step 1 of the process and if I were to print it I wouldn't want a noise reduced version to be used. Unless noise reduction doesn't make much of a difference if it's done before or after the other adjustments? Anyone have an opinion on that?

staereo
11-07-2006, 03:59 PM
Anyone have an opinion on that?

I'm with you. But, as I said before, I take the RAW file, and convert it based on my final product. That means if a client wants online preview, AND prints, I'll end up converting and post processing 2 seperate images all the way through.

Is this convenient? No. But typically I just do the online previews, then reconvert the images that will go to press/printing, and pp just those out of the full batch that went to online previews.

To be honest, unless an image will be used in an online publication or portfolio, any online viewing I treat as 'proofs' and do very little to the image that takes up my time. I then use that time in the conversion and PP of the images that a client or art director chooses as their select images.

In this, you can see that my noise removal would be part of my PP, and be done twice, each time based on my output. :bigtu:

Bruce