Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations Mike Lewis on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

retain photo quality when greatly reduced

Status
Not open for further replies.

rickwells

Technical User
Jul 30, 2003
209
GB
I see some suberb quality graphics on websites but can't acheive this myself.

When I reduce large files down to thumbnail size in photoshop, the quality deteriorates quite dramatically. Yes I do the process in stages and have tried all techniques but no joy. I understand there are some apps out there specifically for this purpose.

Anyone got any experience of these apps. or got expert tips for photoshop.
 
Have you tried using the "Save for Web" feature?

InDesign CS, Photoshop CS, Illustrator CS, Acrobat 7
Windows XP & MAC OS X (Yes I Use Both, Not at the Same Time)
 
Yes - this is my normal practice. Losing quality through optimising for web is not the real problem. It's the reduction of the graphic in P/Shop from, say 18mb, to around 50k. This is where the quality degrades. I would then save for web.
 
Not quite sure what you mean by reducing size in PS from 18mb to 50kb.

I assume that your flattening your pics and then changing resolution to 72 dpi for initial reduction as well as making sure that you're in RGB mode. Are you doing anything else?

One thing you might try is using the Web Photo Gallery (file menu/automate) and select Options/Large Images and Thumbnails as you like (size jpeg quality etc).

If this works to your satisfaction you can put the images generated in web photo gallery in a website design of your choice.

Using OSX 10.3.8 on a G4
 
I mean that I would ( for instance ) take a 19mb 300dpi tiff file CMYK ( 2500px X 2000px )and reduce it to the pixel dimension a require for the web page.
Convert to rgb and to 72 dpi.
Actually, I don't need to flatten or change to rgb or change the resolution - it's all done automatically in "save for web". However I usually do these things first.
I've been using P/Shop commercially for about 9 years and yet I see some picture quality on websites and wonder how this is acheived.
I haven't used this Web Photo Gallery - I'll give it a try - it might throw up some surprizes!
Thanks
 
I would not suggest using save for web or the web photo gallery for two reasons, number one it removes metadata from the images, and if you are a professional that should be plenty of reason. Number two is quality, if you want a quality image why would you want to let someone else tell you how to lower the size of YOUR images.

What are the pixel dimention requirements for the image? How small do you want/need the images?

Part of your problem might be that you are starting with a CMYK image....WHY?? I routinely deliver CMYK images to clients(mostly designers/printers) but I NEVER EVER EVER convert to CMYK until I am COMPLETELY finished with the image. Why limit yourself to such a small color space??

One other tip for higher quality images, when reducing the size of images by going to image, image size, use bicubic smoother, and you can also try reducing your images in steps, like 10% at a time. If all of your images are the same size all the time(an maybe even your web size) just make yourself an action...The computer will do it in 1/10th of the time it would take you! Why not make Photoshop work for you not you work in Photoshop?

Please post a few images so we can look at them. I will gladly do the same so you can see what I do. Can you post a high res so I can get a crack at it? What version of Photoshop are you using? What are your color management settings? Have you ever looked at them (not to be rude, but alot of people, yes even professionals, have never looked at them and do not know what they mean or do...this BLOWS MY MIND!!)

DBX
(Try it my way, it might work...)
 
DBX says "use bicubic smoother" and he may be correct but
I've always used a different set of rules for sizing:

Nearest Neighbor (Jagged) for the fast but less precise method.
This method is recommended for use with illustrations containing non-anti-aliased edges, to preserve hard edges and produce a smaller file. However, this method can result in jagged effects, which become apparent when distorting or scaling an image or performing multiple manipulations on a selection.


Bilinear for a medium-quality method.
It is best to use Bilinear to shrink images


Bicubic (Smooth) slow but more precise resulting in the smoothest tonal gradations.
It estimates the color at a pixel in the destination image by an average of 16 pixels surrounding the closest corresponding pixel in the source image.


It is best to use Bicubic to enlarge images

fact Bicubic is more precise, but only when it comes to enlarging. When it comes to shrinking its exactness can actually produces pixelation, because to shrink an image pixels must be discarded anyway.

This can give acceptable results for up to 2-3 times magnifications. Upsized images become softer, they need to be sharpened with the Unsharp Mask filter.

It estimates the color at a pixel in the destination image by an average of 16 pixels surrounding the closest corresponding pixel in the source image.




The exact numbers will depend on the algorithm used. However, remember that computers work in binary. You will always get the best results by sticking closest to numbers that are related to 2. That is why 75% and 66.666667% work so well (remember, we are talking areas here). Reducing is always much easier because you are discarding information. Enlarging is harder because you are making up information to fill in blanks that aren't there to begin with. So, 110% at a shot sounds like a good place to start. It is a small number so you are only creating information in small steps (consider it like trying to fill in missing map information by increasing the scale. The larger the steps you zoom in, the more information you have to fill in at a time and the less accurate the results) 110% comes out close to one of the factors of 2 for a quadratic interpolation (it may be even better for a bicubic, I don't know).


 
DBX
Thanks for your post - you make some valid points.
Yes often my source image is CMYK but the more recent digital photograhy gives me RGB files.
I have tried reducing file sizes in stages but I'm not happy with the results. Haven't used the bicubic smoother option for recing images - I'll do some tests ( I thought this was for getting best quality when ENLARGING ).
Pixel dimensions - I would need thumbnails at around 100px X 65px. These would link to larger pics of around 350px X 250px.
I'm using Photoshop CS (not CS2) but I have to admit that colour management is something I have not paid enough attention to. You see most of my work has been for litho printing and I have tweaked my set up following proofs and print results to the point where I can be pretty confident of getting the correct results. Not very scientific, I know, but it has worked for me.
The problem now is taking the stuff for print and converting it to very small 72dpi graphics.
Someone told me that there was a stand-alone app which did this job and that was the reason for my original post.
I have never posted any files for viewing on this forum - what is the preferred way to do this and how would I get an 18meg file to you if I were to take up your kind offer of "having a crack at it".
 
Bicubic sharper is used for reductions. If you use it, you gain nothing by going in small increments.
 
Mscallisto
Thanks for all that in-depth stuff. Although I've read it before I must admit I tend not to follow the rules rigidly (probably to my cost) Your reminders of how it all works has set me off to do some more 'scientific' tests with scaling my graphics files.
I still haven't tracked down any app that will do all this for me (speed bonus).
Does such an app. exist, or is this a 'red herring'.

Oh, one thing - could you please explain "...closest to numbers that are related to 2. That is why 75% and 66.666667% work so well..." (excuse my ignorance)
 
You'll have to excuse my ignorance, I read this on the web some years back and believe I posted it in this forum back then.

I played with bicubic for enlarging and bilinear for shrinking as opposed to nearest neighbor that I had always used and liked the results and haven't use nearest neighbor since. I agonized over the " 75% and 66.666667%" and could never figure out what it was all about but again I enjoyed the results.

I'm not saying this is the best method and note that kiddpete says "Bicubic sharper is used for reductions" and since kiddpete has posted some good stuff here I too will have to do some more investigating.
 
From my quick tests last night I would say that bicubic sharper is indeed the best for reductions. I would agree that doing it in stages has no benefit. Thanks to Kiddpete.
I also tried reducing in 2 stages using various permutations of the 5 options in photoshop. Interesting results.
Thanks guys for your input - I won't prolong this debate any further.
If anyone comes across one of these apps which do clever stuff at a click - please let me know!
 
have you tried unsharp mask? This works for me and then save for web.
use "high" jpeg.
 
by reducing the filesize that much you will need to unsharp mask the picture afterwards (just as Kentucky points out)

Kind Regards
Duncan
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top