images just seem to lose there “bang” when I downsize them to put on the web

[steve_johnston9]

<p>Some of my images just seem to lose there “bang” when I downsize them to put on the web. They look fantastic at full screen size, but when I reduce them for the web they lose there bang. Now I understand the files are smaller so lose detail, but other photographer’s website’s seem to be able to retain the sparkle, and display at a size similar to what I wish to display at. I have saved as 12 in terms of JPEG quality in CS4, am I missing something ?</p>

[Matt Laur]

<p>Are you sharpening after you downsample?</p>

[steve_johnston9]

<p><strong>Are you sharpening after you downsample?</strong><br>

No, I take it from the question that I should be doing this. Do you have a recommendation as to how much ?</p>

[Matt Laur]

<p>It really depends on the image itself. I use a light touch with an unsharp mask tool, but sometimes only do so to certain parts of the image. This is, alas, a black art that is highly subjective. But whichever method you use, you'll almost always want to use a (subtle!) bit of it after you take all of that detail away during a down-sampling. And of course, the more you compress the image (the smaller the file, the lower the quality rating, the higher the compression factor chosen during that final JPG save) the less detail that will be preserved. Different images require a different touch in that regard, as well.</p>

[steve_johnston9]

<p>Is there any benefit in fiddly around with the DPI figure ?</p>

[monika_epsefass]

<p>Fiddling around for sharpening? No. For web output, 72 is the rule.</p>

<p>I do an initial sharpening to the full-sized image with smart sharpen and a high ratio (120% or so, radius 0.9, remove Gaussian blur), then downsize via turning dpi from 300 to 72, and set the image to, let's say, 900x600 pix, and sharpen again, this time with a ratio of 40% while the other settings do not change.</p>

[jamila]

Are you changing profile to sRGB to post on the web?

[Brad_]

If you're looking to get some on-screen snap (or "bang" as you put it) in your images from CS, try wide

radius unsharp masking.

 

It's an easy to do trick that can give good results. There can be some

downsides with some images, though, such as noticeable halos. It's similar to positive Clarity in

Lightroom. It boosts local contrast.

 

Try Unsharp Mask (20, 50, 0) and adjust the first parameter (Amount) to suit. I sometimes use that with

Amount set to 10-15 after I export images from LR that will end up on the web.

 

>>> For web output, 72 is the rule.

 

DPI is irrelevant for web output. Only pixel dimensions matter.

[Tim_Lookingbill]

<p>No single one size fits all sharpening recipe is going to work here.</p>

<p>Use your eyes viewing the downsized final at 100% zoom view on a calibrated and profiled display in a color managed application. The sharpening settings will be different on an image with a lot of high frequency textures like trees and foliage in a landscape compared to images of custom hot rod cars with pearl luster paint jobs. </p>

[drjoder]

<p>Steve, I think Jamila has a pretty likely answer. Confirm that when you do the downsizing that you are indeed retaining (or changing to) the sRGB color space. If not, your color images will look kind of dull. I ran into this when I first started posting on photo.net as I was wrongly posting my images in the Adobe color space.</p>

[alan_olander1664878205]

<p>72 ppi is not the rule. PPI makes no difference for web posting, only printing.</p>

[patricklavoie]

<p>i personnaly think that you can apply the same sharpening to all your smaller images, according they have been correctly work in there original size. Have a look at my web site.. when i create my smaller web version from the hirez, i dont have time to open each of them and fiddle around to find the magic number; pixel size + srgb colorspace + jpeg quality 8-10 + a good amount of sharpening (one that you feel comfortable, with the method you feel comfortable).</p>

<p>The USM trick that Brad talk about is a excellent method for local contrast enhencement, not the same as a sharpen, but give more depth to a image. This should have been done on your hirez .. if you never done that, your image are missing something even in there hirez state ; )</p>

<p>Basically, if your original hirez is really good looking, just make sure they are in srgb, and apply them your sharpening .. same amount to all of them.</p>

<p>I have a create a action in Photoshop to do that for me, and use it with Image Processor.</p>

<p>* if you use the bicubic sharper as your default image size setting, i think it should apply that when it create the smaller version, and if so, using this method, a sharpen is already apply during the process... will need to verify that, but im pretty much convince about that.</p>

[KenPapai]

<p>Correct Patrick -- when you downsize and Bicubic Sharper is on, then little to no need to further Smart Sharpen. <br /> Maybe a slight USM would help -- e.g., Patrick (and now I) uses 15-15-1 nearly always.</p>

[lachaine]

<p>One thing you could do, is to save the final version from whatever editor you are using without downsizing it. Then use an image viewer to do the final web-sized version. This could be something like Irfanview, Xnview, Faststone Viewer (all free by the way, and easy to work with). Open that large version with the viewer, resize it to whatever small size you want using the default resampling method (bicubic or lanczos). Once resized, still in the same viewer (before you save the file), apply a little sharpening to it (it doesn't take much at that size, maybe a value like 1 or 2 depending on the software and how sharpened it was to begin with). Sometimes, an image can use a little enhancement of contrast as well, before the sharpening). Then save it as the final web-sized JPEG. I like to use Faststone Viewer for this, because this viewer makes it so easy to experiment with the image and see exactly what you're going to get as it will be seen on the web - before you actually do it.</p>

<p>Regardless of the above, even if you do it all in your original image editor, I've always found that a large image can often benefit from slightly increased contrast (or even local contrast enhancement) once resized small, and then a little sharpening. This is what will make the image pop.</p>

[monika_epsefass]

<blockquote>

<p>72 ppi is not the rule. PPI makes no difference for web posting, only printing.</p>

</blockquote>

<p>Why is it that everybody tells me images for the web should be 72 ppi? Is that one of the urban myths? In the beginning, I've uploaded any ppi, and nobody ever noticed. ;-)<br>

Could someone clarify this, then? Thanks.</p>

[Matt Laur]

<blockquote>

<p>Why is it that everybody tells me images for the web should be 72 ppi?</p>

</blockquote>

<p>People who tell you this simply don't understand what they're talking about, that's all. If you think about it, you can set your editing software to make the image 72ppi, but ... then what? You still have to provide some sort of dimensions. When you set up an image for display on a screen, you're thinking in terms of pixels. 800 pixels wide. 600 pixels tall, etc. That's it. The web browser or other application that's rendering the image on your screen doesn't think in terms of inches or centimeters ... it just takes up, for example, 800 by 600 pixels worth of the screen's available pixels, and displays the image. I'm simplifying a bit, because some applications will scale the image down to make it fit if it's too big, but you have no control over that.<br /><br />The dpi/ppi value that's recorded in the header of that JPG file doesn't in any way change the number of actual pixels that make up the image file. And it's that number of of pixels that dictate how big the image is on the computer screen. If you want the image to be 700 pixels wide on the screen, make it 700 pixels wide. That's all there is to it. You could have the dpi value set to 1 or 1000 and it won't change the fact that it's 700 pixels wide.<br /><br />The only time ppi comes into play is when certain applications use it decide how large the image should be (how may pixels should be spread out over how many inches of paper) when it's <em>printed</em>. And even then, printer drivers and other software can scale on the fly anyway. </p>

[monika_epsefass]

<p>And that's exactly how I work, Matt. Maybe my initial line was misunderstanding - I've so often heard that monitors display with a 72ppi-ratio that common standard is to scale to 72pix and then determine the overall size of the image.<br>

I love printing much better, anyway. ;-)</p>

[KenPapai]

<p>That's not correct either/ 72 ppi for monitors is just an "old standard" that means little to nothing. <br>

Do the basic math yourself:</p>

<p>How wide is your monitor in inches? What is its native resolution (most common is 1280 pixels wide). Doing the division tells you your monitors native "PPI". That varies from 70 up to 110.</p>

[Tim_Lookingbill]

<p>Monika, I hope you don't include me as among the everyone who says to size to 72ppi even though I've often said this in forum threads in the past due to my lack of visualizing mathematically described dimensional demarcation within the cyber world that can't be physically seen. It is a common mistake I make constantly even after researching and practicing digital imaging for 12 years.</p>

<p>When someone describes physical dimensions in the real world as I did before digital with a graphics camera and proportion wheel, no problem. Describe it referenced inside a tiny box that flashes lights and pretty colors mathematically at you and asks you to say..."See! It's this big!" and I get lost.</p>

<p>I know I have to view and describe resolution to something physical like comparing 200-1000 thread count king sized sheet. Even though the percale can change, the size of the King size sheet stays the same. Saying just 72ppi or thread count is like saying I have a 1000 percale sheet and not mentioning whether it's twin, full, queen or king size.</p>

<p>Why can't I just walk away from these discussions?</p>

[john_markanich]

<p>Are you sure that YOUR monitor is properly calibrated to begin with? (using one of the commercial pucks available). No whiz here, just a thought. </p>

[dave_wilson1]

<p>Reading this thread tells me that I don't have a @#$% clue what I'm doing on the computer in post work... My JPEGS where I set everything in the camera look great, my RAW conversions all stink. I'm not doing any of this, but I have tried the unsharp mask as described in Scott Kelby's older book and it works some.</p>

[dave_wilson1]

<p>So I just spent three hours messing with a few RAW files and converting to JPEG and using some of these recommendations. I see improvement for sure, but still not where I want it to be so I have to try some other combinations, but I see where if you don't adjust each size like Matt indicates it's worse each time. Very good question Steve 8-)) I also thought some of my sharpening numbers lined up with what Brad said without looking at them. So I'll keep trying.</p>

[robert_k1]

<p>OP would have gotten much better responses if he provides:</p>

<p>- the pixel sizes of the full screen and the downsized<br>

- how is the downsize done?<br>

- what "bang" is lost in the downsized? In tonal range, in colors, in sharpness?<br>

- what kind of image? With lost of fine details, or with very little?</p>

<p>Each of the above can point to a different problem.</p>

[Tim_Lookingbill]

<p>Agreed, Robert.</p>

<p>I went back to the original OP's post and I think Matt with all good intentions got the ball rolling on sharpening after downsizing in determining the solution to the loss of "Bang".</p>

<p>I don't have an image quality issue after downsizing on my 6MP Raws even after uploading to Photo.net or Flickr. I just zoom down to 25% view on the full resolution version in PS to see how it looks downsized and get a very close match to actual downsizing to 700 pixels on the long end at 72ppi using Bicubic Smoother and then using Save For Web at 50-60 Quality level to save to jpeg.</p>

[indraneel]

<p>You might want to take a look at this page:<br /> <a href="http://www.4p8.com/eric.brasseur/gamma.html">http://www.4p8.com/eric.brasseur/gamma.html</a><br /> From what (little) I understood, downsizing in non linear gamma seems to play an important role. Few editors perform resampling in a gamma correct way. This leads to loss of details in fine contrast. Solutions are listed in the article.<br /> Briefly, a workaround is to convert your image to a linear colorspace, resize, and then convert back to whatever you want. 32bit/channel is preferred. Linear colorspaces can be made with photoshop.</p>

<p> </p>