Compression Choices… what is really noticeable?

[Don]

As a new member of the PTE world, I have learned much about this wonderful software from this forum and the tutorials. One issue that I have not been able to clearly understand has to do with file compression methods. I’ve read various viewpoints here, but still didn’t think that I had a satisfactory answer for myself regarding what I should do.

So, I thought that the best approach was for me to do some primary research. I took an image that I shot and saved it in various ways, put it into a PTE slideshow, and posted it at MediaFire. The shot I chose was taken with a Canon 40D and I did some minor Photoshop work on it. It was originally shot as a large JPEG, as I was not yet using RAW at the time I took this picture (but I am now.) Each version of the image was saved only once, from the PSD file. As you can imagine, the file sizes varied considerably, from 3078 KB when the JPEG was saved using a 12 compression down to a 282 KB at a 0 compression.

I really expected to see a more dramatic difference. Am I missing something? I’m viewing this on a 24” LCD (1920 x 1200.) Does this mean that saving at higher compression factors is only wasting space? Possibly I should have used an image with a sharper focus to bring out any critical differences. This was hand-held, 100 ISO, f/4.0, 1/160 sec, with a 55m focal length.

I would like some opinions on this comparison. Here is the MediaFire link:

http://www.mediafire.com/file/y4mymznmomm/File Compression Comparisons.zip

The slideshow is 1 minute 52 seconds and the file is 17.29 MB zipped.

Thanks,

Don

[fh1805]

Don,

You could have built the sequence to be more "user friendly" for the comparison. I would have loved to have manually switched to and fro between the adjacent JPEG12 and JPEG6 images but couldn't. Although I could pause the show with the spacebar and then use the arrow keys to move forward and back, this technique skipped over the adjacent JPG12 and JPG6 images. I was taken from the screen of text about them straight to the 4-up and back again.

I didn't notice too much difference between any of the images. But then I'm not a hyper-critical person. I didn't find the 4-up particularly useful.

I'm the coordinator of our local A-V Group and it's our meeting tomorrow evening. I'll ask the members what settings they use and report back sometime over the weekend.

regards,

Peter

[Lin Evans]

Hi Don,

First, let me explain the "numbers" used for compression. I'm assuming your are using Photoshop or a similar photo editing program since you mentioned 12.

The higher the number, the "less" compression. A Photoshop 12 compressed jpg file has very little compression. Any number below 4 is very highly compressed. High compression with jpg images can leave artifacts which will become especially visible when the image is displayed at full size such as during deep zooms.

Compression is only a storage media space saver. Once the file is displayed on a computer, the image is uncompressed so that it expands to its full size just as an uncompressed tif (tagged image format) file. All jpg's are 8 bit so they expand to the same size as an 8 bit tif file.

Typically, if you use nothing greater (no lower number) than compression level 8 in Photoshop, there will be no significant "visible" difference in your slideshow even if you do deep zooms to and beyond 100% of the full size of your image. Some are happy with even a compression level of 6 or 7, but personally I don't use a higher compression than 8 myself.

If you are using full sized files from your camera in your slideshow, they will be downsampled considerably for most presently available displays. The highest resolution computer monitor currently available is about 9.2 megapixels, so even were you displaying your 40D (10 megapixel) images on such a system, they would be downsampled to a degree. There are not very many of these and they take special video cards (usually two) to produce this resolution. The next best monitors produce a 4 megapixel resolution (2560x1600) and few have these. The most common higher resolution displays are 1920x1200 pixels with 1920x1080 resolution becoming quite common to match the 1080p HD televisions. So essentially, when use images at their full original pixel dimensions in your slideshow, they will be downsampled considerably. The process of doing this can ameliorate some of the artifacts of compression which would otherwise be visible were you to zoom to the original pixel dimensions for displaying a part of your image.

Where you really see significant compression artifacts is when you compress an already downsampled file greatly. If you were to downsample your originals to about 1024x683 pixels then compress them using 4 or 5 in Photoshop and then zoom in on an area you will quickly see the jpg artifacts, especially if you have an LCD display or a really sharp CRT display.

The thing to keep in mind is that you damage your images when you use great compression (low numbers) so always work with a copy and preserve your originals.

Best regards,

Lin

[Conflow]

DON,

Following up from Lins' and Peters' replies and considering the inability of most Members trying to make a comparison

between the Images you posted on 'Media-Fire' due to size comparison limitations, perhaps a more fundamental

understanding of the JPeg-Process would benefit yourself and other readers.

One immediate problem encountered was the 12:1 Compression-ratio you used as an example because 'Photoshop'

considers that to be the maximum compression ratio which can be used before small barely perceivable artifacts

become evident in any edited Image ~ consequently the example's were not the best choice's for discussion.

I "attach" a PDF Document explaining not only the Image degradation factors of compression but all the other more

serious things like 'Text-Captions' which JPegs hate, as well as 'non-ratiometric' cropping etc,etc.

Hope this gives a better understanding as to how far we can go with Compression and other Editing factors as there

are many excellent illustrations within the Pdf-File..

Brian (Conflow)

(Images from Photo-Britannica)

JPeg-Compression.pdf

[Lin Evans]

Hi Don,

To augment Brian's example, here is a link to a quick slideshow and a couple sample images.

I took a quick jpeg snap of a motor home with a 40D. I saved the jpg at Photoshop's lowest compression level 12. I then saved the original again at Photoshop's compression level 3. Next, I downsampled both the level 12 and level 3 to 1024x683 pixels then used PhotoZoom Pro to upsample both to 12000x8000 pixels. I cropped a like section of the vehicle license plate from the level 3 and level 12 compressionsand placed those in the slideshow so you can see the effects of high compression with a sample from your own model camera.

I include links to the original interpolated crops:

http://www.learntoma...compression.zip (about 6.5 meg Windows exe)

http://www.learntoma...ion/level12.jpg

http://www.learntoma...sion/level3.jpg

Best regards,

Lin

[Conflow]

Hi Lin,

Your Mini-Slideshow is a great 'dynamic-example' of JPeg-Compression artifacts.

I guess the old saying is true:-.."Photo-examples are worth a thousand words"

I do hope readers will download my Pdf and your Mini-Slideshow for reference,

because what is written in the Documentation is well demonstated in the Mini-Show.

Brian (Conflow).

[Don]

Don,

You could have built the sequence to be more "user friendly" for the comparison. I would have loved to have manually switched to and fro between the adjacent JPEG12 and JPEG6 images but couldn't. Although I could pause the show with the spacebar and then use the arrow keys to move forward and back, this technique skipped over the adjacent JPG12 and JPG6 images. I was taken from the screen of text about them straight to the 4-up and back again.

I didn't notice too much difference between any of the images. But then I'm not a hyper-critical person. I didn't find the 4-up particularly useful.

I'm the coordinator of our local A-V Group and it's our meeting tomorrow evening. I'll ask the members what settings they use and report back sometime over the weekend.

regards,

Peter

Hi Peter,

Sorry about the “user-unfriendly” nature of the comparison. You are right, it would be easier for someone to evaluate if I had built a bit more flexibility into it. Another mistake I made was to use O&A for the JPG12 vs JPG6 comparison, which likely accounts for you not being able to navigate back and forth between them. I’m still quite new with PTE, so please bear with me. I value the discussions that I read daily in these forums. They are helping me, and I’m sure others.

My intent here was to determine the best compression level for me to use when saving images to JPG for PTE. In my non-PTE photography, I use PSD unless I’m sending the images to someone that does not have Photoshop. In that case I save the images as JPGs with the least possible compression (12.)

The file sizes for the initial slideshows that I’ve made were much too large, especially when compared to those that I’m seeing on this and other sites. My sense is that this not only makes distributing them on the net inefficient, but also degrades the viewing experience if another person is using a computer with less capability. So, I’ve been exploring ways to get my file sizes down. My first step was to reduce the pixel dimensions from the 3888 x 2592 that my camera captures to 1920 x 1200, which is my monitor resolution. That helped a lot, but I thought that compressing the images more might further improve the situation. As most of my efforts had previously been directed towards printing photos, my predisposition was to use as little compression as possible. The experiment that I posted here was done to challenge my concept of “less compression is always better.” Evidently, that is not true.

I eagerly await the results of your discussion at your local A-V group.

Thanks,

Don

[Don]

Hi Don,

First, let me explain the "numbers" used for compression. I'm assuming your are using Photoshop or a similar photo editing program since you mentioned 12.

The higher the number, the "less" compression. A Photoshop 12 compressed jpg file has very little compression. Any number below 4 is very highly compressed. High compression with jpg images can leave artifacts which will become especially visible when the image is displayed at full size such as during deep zooms.

Compression is only a storage media space saver. Once the file is displayed on a computer, the image is uncompressed so that it expands to its full size just as an uncompressed tif (tagged image format) file. All jpg's are 8 bit so they expand to the same size as an 8 bit tif file.

Typically, if you use nothing greater (no lower number) than compression level 8 in Photoshop, there will be no significant "visible" difference in your slideshow even if you do deep zooms to and beyond 100% of the full size of your image. Some are happy with even a compression level of 6 or 7, but personally I don't use a higher compression than 8 myself.

If you are using full sized files from your camera in your slideshow, they will be downsampled considerably for most presently available displays. The highest resolution computer monitor currently available is about 9.2 megapixels, so even were you displaying your 40D (10 megapixel) images on such a system, they would be downsampled to a degree. There are not very many of these and they take special video cards (usually two) to produce this resolution. The next best monitors produce a 4 megapixel resolution (2560x1600) and few have these. The most common higher resolution displays are 1920x1200 pixels with 1920x1080 resolution becoming quite common to match the 1080p HD televisions. So essentially, when use images at their full original pixel dimensions in your slideshow, they will be downsampled considerably. The process of doing this can ameliorate some of the artifacts of compression which would otherwise be visible were you to zoom to the original pixel dimensions for displaying a part of your image.

Where you really see significant compression artifacts is when you compress an already downsampled file greatly. If you were to downsample your originals to about 1024x683 pixels then compress them using 4 or 5 in Photoshop and then zoom in on an area you will quickly see the jpg artifacts, especially if you have an LCD display or a really sharp CRT display.

The thing to keep in mind is that you damage your images when you use great compression (low numbers) so always work with a copy and preserve your originals.

Best regards,

Lin

Hi Lin,

Thanks for the response. I had not considered the impact of deep zooms, which is obviously a good point. I have been experimenting with deep zooms recently and clearly see image quality issues. More compression might complicate this.

Separately, I have viewed a video of yours that introduces the concept of shooting the same subject at progressively higher settings of a telephoto lens (deeper zooms, that is.) As your demonstration showed, this can preserve image quality while allowing for deep zooms at high quality in PTE. If others have not seen this, I suggest that they do.

As I mentioned in my response above to Peter, I do downsample my image sizes prior to saving them as JPGs. After editing in Photoshop, I crop to a 1.6 : 1.0 ratio by setting the crop tool to 10 in by 6.25 in. Then I downsample by clicking Image>Image Size and enter 1920 as the width for a landscape image, leaving Scale Styles, Constrain Proportions, and Resample Image (Bicubic) all checked. The height automatically changes to 1200, as this is linked to width. If the image is portrait orientation, I set height to 1200 and let the width fall as it may. I leave the resolution alone, since I’m assuming that this does not affect images shown on a monitor. Do you have any suggestions about this about this approach? I used this method when I prepared the test that I posted on MediaFire.

Thanks,

Don

[Don]

DON,

Following up from Lins' and Peters' replies and considering the inability of most Members trying to make a comparison

between the Images you posted on 'Media-Fire' due to size comparison limitations, perhaps a more fundamental

understanding of the JPeg-Process would benefit yourself and other readers.

One immediate problem encountered was the 12:1 Compression-ratio you used as an example because 'Photoshop'

considers that to be the maximum compression ratio which can be used before small barely perceivable artifacts

become evident in any edited Image ~ consequently the example's were not the best choice's for discussion.

I "attach" a PDF Document explaining not only the Image degradation factors of compression but all the other more

serious things like 'Text-Captions' which JPegs hate, as well as 'non-ratiometric' cropping etc,etc.

Hope this gives a better understanding as to how far we can go with Compression and other Editing factors as there

are many excellent illustrations within the Pdf-File..

Brian (Conflow)

(Images from Photo-Britannica)

Hi Brian,

Thanks for the PFD file. It provides a rather complete explanation. I’ve finished a first reading, but will need to spend more time with it prior to gaining a full understanding. The visual examples are extremely helpful. Among other things, I was not aware that JPGs did not handle text or line drawings as well as images with smooth variations. There is much to learn from this document.

Thanks for posting it.

Don

[Don]

Hi Don,

To augment Brian's example, here is a link to a quick slideshow and a couple sample images.

I took a quick jpeg snap of a motor home with a 40D. I saved the jpg at Photoshop's lowest compression level 12. I then saved the original again at Photoshop's compression level 3. Next, I downsampled both the level 12 and level 3 to 1024x683 pixels then used PhotoZoom Pro to upsample both to 12000x8000 pixels. I cropped a like section of the vehicle license plate from the level 3 and level 12 compressionsand placed those in the slideshow so you can see the effects of high compression with a sample from your own model camera.

I include links to the original interpolated crops:

http://www.learntoma...compression.zip (about 6.5 meg Windows exe)

http://www.learntoma...ion/level12.jpg

http://www.learntoma...sion/level3.jpg

Best regards,

Lin

Hi Lin,

Your test does clearly show a noticeable difference.

Thanks for your efforts and help.

Don

[Lin Evans]

Hi Don,

Essentially, image quality will not be enhanced by having a greater pixel count in your files on the slide list in PTE than the display device which the slideshow will be viewed "except" in the aforementioned example of deep zooms. The way I handle it (deep zooms) is to only use higher pixel count (resolution) slides where I will be doing zooms to reveal detail, etc., and then I generally start the zoom with an image of normal display resolution and duplicate my settings for a "quick" "no transition" into a "crop" of a second slide which constitutes a high resolution duplicate of that portion of the original slide. This saves some storage space and is easier on video cards which are less than optimal.

The way you are preparing your slides seems logical and appropriate to me. You can also just do your pseudo "crops" right in PTE by simply adjusting the position of the larger image to duplicate what you achieve by doing a physical crop in Photoshop.

One of my pet "annoyances" is that the movie industry has apparently dictated that we use a 16:9 aspect ratio which is counter intuitive for the photographer who has many dollars invested in 35mm format equipment. It forces us to shoot wide and waste "pixels" to accommodate the later need to crop for display on HD TV or even on commonly produced LED/LCD monitors which are increasingly being produced at 1080p (1920x1080). Even though some of the newer dSLR's allow for direct capture at at 16:9 aspect ratio, this wastes valuable sensor resolution. I suppose as higher resolution sensors become commonplace, the damage is less problematic, but it is interesting to see that the very "best" movie format (in my opinion) IMAX 3D, is still much more similar to our 3:2 than to 16:9. Having numerous digital cameras of both professional and enthusiast level dSLR and digicam type, I just hate to have to change my shooting style which has been habitual since 35mm format was first available to one which forces me to shoot wider than I like so that I can "crop" to accommodate 16:9 aspect ratio.

The important thing, in my opinion, is to use high enough resolution on the slide list but not to overdo. In the recent past I have created slideshows for some of my art gallery clients who purchased 9 megapixel displays. These were wonderful for non animated slideshows, but were incredibly expensive (over $6,000 U.S.D. plus another $2K for video cards). Now most of them have gone to the 30" 4 megapixel displays which can be had for under $2,000 and can be run with a number of single reasonably powerful video cards. The Apple Cinema 30" 2560x1600 pixel display is quite popular and our PTE shows are striking on this display which retails for about $1700. For shows on these displays, I find it necessary to use at least four megapixel files and for deep zooms I use full resolution images from my cameras and sometimes interpolate with PhotoZoom Pro 3 to get it right for deep zooms.

As for actual compression, I doubt that you will see any real differences between level 12 and level 6 Photoshop compression in most cases. For those few slides where you do deep zooms, I would keep it at at least Photoshop 8 especially if there is any text involved.

Best regards,

Lin

[fh1805]

...I'll ask the members what settings they use and report back sometime over the weekend...

Sorry for the couple of days delay in getting back. I did ask our members what they used and the results were as follows:

Level 12 - seven people

Level 10 - four people

Level 9 - three people

Level 8 - six people

Level 6 - two people

and one member said he didn't use JPEG levels but used "Save for web" and aimed for a particular file size. He said this option gave him a range of compression between 0% and 100%.

I don't know whether that helps inform the debate, but I promised to report back - and now I have done.

regards,

Peter

P.S. If it's of any interest, I'm one of the six using level 8.

[LumenLux]

Hi Don,

The way you are preparing your slides seems logical and appropriate to me. You can also just do your pseudo "crops" right in PTE by simply adjusting the position of the larger image to duplicate what you achieve by doing a physical crop in Photoshop.

Best regards,

Lin

Lin, you, and this topic contain a wealth of useful info. I agree with both your tech analysis and your sentiment on monitor/sensor ratios. I quote above, the one statement where I may not understand your intent. If Don is attempting to not oversize his images and file sizes, is there any reason to "adjusting the position" and thus keep the larger file size? Or are you suggesting this as a convenience with little sacrifice via the larger file?

[Lin Evans]

Hi Robert,

Yes, just a convenience. Sometimes, when the original file size being used is not too large so as to place a burden on the hardware, I find it easier to accommodate the 1920x1080 wide screen by just positioning the image to encompass the same geography as it would have contained had I physically cropped the image in Photoshop,

Perhaps this habit of mine is less useful to some. I have a number of clients who use 2500x1600 pixel displays. I often make a copy of my originals sized at 2500x1667 and then simply adjust the position to slightly crop the vertical aspect to fit the 2500x1600. I just adjust the zoom in PTE for these essentially four megapixel images and "pseudo crop" by positioning for 1080p rather than again manipulate the images thus having a third set of essentially the same captures.

I also frequently shoot my wildlife using a Sigma SD14 and sometimes an SD10 where the native file sizes are 2640x1760 or 2268x1512 respectively so that there is only a small amount of zoom either way to fill a 4:3 aspect ratio or to accommodate a 1920x1080 without reworking the originals at all.

Best regards,

Lin

Lin, you, and this topic contain a wealth of useful info. I agree with both your tech analysis and your sentiment on monitor/sensor ratios. I quote above, the one statement where I may not understand your intent. If Don is attempting to not oversize his images and file sizes, is there any reason to "adjusting the position" and thus keep the larger file size? Or are you suggesting this as a convenience with little sacrifice via the larger file?

[Don]

Sorry for the couple of days delay in getting back. I did ask our members what they used and the results were as follows:

Level 12 - seven people

Level 10 - four people

Level 9 - three people

Level 8 - six people

Level 6 - two people

and one member said he didn't use JPEG levels but used "Save for web" and aimed for a particular file size. He said this option gave him a range of compression between 0% and 100%.

I don't know whether that helps inform the debate, but I promised to report back - and now I have done.

regards,

Peter

P.S. If it's of any interest, I'm one of the six using level 8.

Peter,

Thanks for the info. I had not considered "Save for Web." I should try that.

Don

[LumenLux]

Peter,

Thanks for the info. I had not considered "Save for Web." I should try that.

Don

I often use Save for web as I think it yields excellent quality. But does someone have an automated routine for using it for batch processing many images for PTE?

[LumenLux]

You can either create a photoshop droplet or use software that is designed just for this purpose.

Adobe Save for Web Video Tutorial

Easy Thumbnails Free Software

Irfanview and many other free image editors also can resize/save jpeg images. Irfanview

tutorial.

Tom

Thanks Tom, I'll look at these options. For the most part, I currently do use Irfanview for my batch sizing, but am specifically interested in batch process via PS or PS Elements.

[Don]

You can either create a photoshop droplet or use software that is designed just for this purpose.

Adobe Save for Web Video Tutorial

Easy Thumbnails Free Software

Irfanview and many other free image editors also can resize/save jpeg images. Irfanview

tutorial.

Tom

Tom,

Thanks. Excellent tutorial on Save For Web.

Don

[KyDan]

Can you stand to read one more opinion

I freely admit that my P2E useage is sporadic at best. Once spring arrives,

my time spent in front of the computer is greatly reduced.

Also, since I am more than a bit frugal, I try to econimize with new

software where ever possible.

Having said all that-- I'm surprised that no one here has mentioned

the free photo editing software from Google.com called

Picasa.

I know it's not in the same league as Adobe Photo shop but

it will allow cropping, re-sizing and some photo editing.

It's very simple to batch re-size an entire folder of images.

For my most recent photo show I re-sized the images mostly for file size

so the resultant exe file would be more "downloadable".

I experimented and found that I could reduce my original photos

from 2272 x 1704 @ around 1 Meg

down to 1200 x 900 @ around .1 Meg (100 k)

with only slight mottling visible in the blue sky shots.

For now at least, I'm using Google picasa.

It's got a fairly easy learning curve and you can't complain about the

price!

My 2¢