Menu Close

Resizing images: Image Size, Canvas Size, Sharpening

Even if you don’t need precise alignment, which happens sometimes, you still have to resize your images in Photoshop. Most of the time model images and close-ups don’t have to be aligned to fit any margins, but you should remember about the proper image size and proportions. It sounds simple. We have this “Image size” command in Photoshop, why not just go there and put some values in, and then just record this as an action and repeat on any other images in need of resizing? Well, there are some things you have to know about resizing first.

Image size

Photoshop has gone a long way with its interpolation algorithms. Every new version deals with resampling better and better. When I reduced images size in CS5, I've had to sharpen them quite significantly, and if I use the same settings when I resize in CC2018, I'll get them critically oversharpened. It means that images get less blurry when they are reduced in size, and that's great.

It might look simple enough, but to resize an image Photoshop has to recalculate every single pixel and resample the resulting image. To do so, it can use different interpolation algorithms you can choose from. One of the bicubic algorithms will be there by default, but if you pick another one from the drop-out menu, next time you use the “Image Size” panel it will be active. What’s the difference? Well, most of them are pretty harmless except one, which is the “Nearest neighbor”. You wouldn’t want it to resize your images, trust me, as they become ugly and the edges look too sharp and jagged. As we’re never going to enlarge images, only make them smaller in size, a good choice would be the “Bicubic Sharper” which, as Photoshop claims, is best for reduction.

But is the "Bicubic Sharper" algorithm really best for catalogue retouching, where we have to reduce image size all the time? It depends. I used it for several years before I even started to have second thoughts about that. And only recently I've made the decision to switch to the regular "Bicubic" algorithm. Here's why. When you resize images with the sharper algorithm, they are sharper indeed, but that might lead to interpolation moire issues. No one wants interpolation moire, trust me. On the other hand, the "Bicubic" algorithm is not as sharp, but it also doesn't give you as much moire. When your images are lacking sharpness just a little bit, you can sharpen them right after resizing with another action, that's easy. But if you get moire, well, it's a bit more of a problem. That's why I switched to the "Bicubic" algorithm. However, I don't claim that it's a perfect choice. It's up to you which interpolation algorithm to use (as long as it’s not the “Nearest neighbor” you’re good).

Some contents or functionalities here are not available due to your cookie preferences!

This happens because the functionality/content marked as “Google Youtube” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.

Let’s give it a test. I have this model image which I cropped in ACR without any particular proportions. I need it to be 1500 pixels high and 1000 pixels wide. How can I do it? When I access the Image size panel and put 1000 pixels as my desired width, it recalculates the height according to the current proportions of the image, and it gives me a value that is less than the one I need. And I can’t edit both values freely as they are linked unless I click on the chain image to unlock this connection. But if I put 1000 in the width and 1500 in the height input fields respectively, and press OK, I’ll get a distorted image. It happens because the initial proportions of the image were different, so in cases like this Photoshop has to stretch or contract the image to make it fit the values entered by the user. And this is certainly is not a good way to resize catalogue images. Don't get me wrong, I'm not against the Image Size command, but it only should be used in combination with other commands, and you'll learn about it a bit later.

Open

Crop Tool

Are there any other ways to resize? Sure, there are. You can also use the Crop Tool, as it is capable of not just fitting the image to some proportions, but resizing it as well. To resize an image with the Crop Tool, you have to specify its desired width, height and resolution. But there’s one thing about the Crop Tool that makes it rather unpredictable, so instead of explaining how to crop and resize images with its use, I’ll show how easily it can ruin your images. When you want to rotate the image and crop and resize it at the same time, the Crop Tool fails to do so. It brings moire to some textures, so they appear rippled. It happens often enough when resizing, but the Crop Tool makes it so much worse, that I have to warn you against its usage for resizing at all. Let’s give it a test now.

I used the Crop Tool to resize the left image. Click on it and look at all the moire
Some contents or functionalities here are not available due to your cookie preferences!

This happens because the functionality/content marked as “Google Youtube” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.

I have this close-up image, which is moire prone because of its small texture. I'll crop both images with a bit of rotation, but for the first copy of the image, I’ll use the Crop Tool set to a ratio, not to a specific size in pixels. It will only crop it, but the size will remain the same, only proportions will change. After cropping I’ll change the image size manually, apply the changes and set the image scale to 100% to see it as it is.

As for the other copy of the same image, I’ll use the Crop Tool with the specific image size, not a ratio, so it will resize as well as crop it. After applying the changes and setting the image scale to 100%, I’ll bring the other copy of the image to compare them. Now, look – one of the images looks defect, as the texture of the jacket became rippled. The other image, which I resized myself, looks normal. The problem is that the Crop Tool uses some weird interpolation algorithm, which can cause such unpleasant effects, and you can’t do anything about it. To avoid it, set the Crop Tool in such a way that it doesn’t resize anything, set it to crop to a Ratio, not to a specific width and height. Otherwise, it might make your images look like this. I have to admit that it doesn’t happen often, as it only happens with specific kinds of surfaces like jackets, sweaters and other knitted garments, but there are other ways of resizing, so why using this faulty algorithm at all?

Open

Image resizing

As a person that is often responsible for the actions of many people with different levels of responsibility, I grew to like foolproof solutions to all the problems I ever face. When you have 25 retouchers under you command and you tell them that all the images should be, let's say 1000 by 1500 pixels, do you really expect them all to comply all the time? Actually, almost every folder they will be passing to you for QA, will probably contain an odd image or two. You can tell them to check, you can fine them if they don't, but still, faulty images will keep emerging. We're all humans after all.

But that doesn't mean I would agree to sit there and catch these faulty images to prevent them from leaking to the website. I'm also a human by the way. But I am a human with a basic knowledge of algorithms and programming. To make sure that the resulting images are of the right size and not stretched or anything, I use quite a few mechanisms. I never resize with the Crop Tool, as it cannot properly rotate and resize at the same time. As I’ve mentioned in the Actions & Scripts section, I use scripts to prevent the user from saving images if they were not resized correctly, I also use scripts to align images. Now it’s time to mention that actions should also be used for resizing of the images that do not need any alignment. It’s really easy to write such an action, as it consists of two simple steps.

Some contents or functionalities here are not available due to your cookie preferences!

This happens because the functionality/content marked as “Google Youtube” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.

The first step is the Image size, where I make sure that no weird algorithms like the Nearest Neighbor are used. As we’re only going to reduce our images size and never enlarge them, the Automatic algorithm is fine, as in such cases Photoshop uses the Bicubic Sharper. Then it’s essential to make sure that proportions are constrained, as it prevents random stretching of images. After making sure that the resolution is set to a desired value, I’ll also set the height. But I’d like to say a few words about resolution first. Since we’re not going to print the images in question, resolution is not really important, as the only thing that matters is the size in pixels. But a lot of people think it’s important and they demand it to be set to some particular values like 72, and I don’t see why we shouldn’t. We’re going to make images with some resolution anyway, so it’s only reasonable if we stick to some value, whatever it might be.

Now I can press OK and apply the changes. The image size will be reduced and the image will reach the desired height. But what about the width? If the image was cropped with a correct ratio before, during the ACR stage or later in Photoshop, it would be automatically set to the right value, too. But if the proportions were off, it would be wrong. So our next step is to ensure that the width is right.

The best way to do it is to use the Canvas Size panel. The only thing to do here is make sure that the Relative checkbox is off and put the desired width value in the width input field. Now we can apply the changes. If the image was wider than necessary, it would be trimmed from the sides. If it was too narrow, it would have its canvas extended and filled with white. It’s not good, but better than a stretched image. Some people use bright colors like red or green to fill the canvas, as it makes it more noticeable, but I’d vote against that. Bright colors make your eyes tired faster, and if you miss a white stripe it’s not such a big deal. But a bright red stripe on the website is a fail for sure. If you wonder how something like a bright red stripe can be missed during the retouching work, I assure you that when you handle hundreds of images daily it’s really easy to miss anything, no matter how crazy it looks. So the color you use for the canvas extension should be noticeable and it should be different from the Photoshop background, otherwise, you won’t be able to see it at all.

Open

In general, to prevent your canvas from extending and giving you more work to do, it’s reasonable to crop the images with a proper ratio or a wider ratio, so that the canvas only get trimmed from the sides. For example, if your images are supposed to be 1000 by 1500, it's reasonable to Image Size them to 1001 in height, and then Canvas size them to 1000 by 1500. This way you'll be sure nothing ever goes wrong. No odd pixels or one-pixel-wide odd-colored lines anywhere.

But sometimes there’s no choice. It happens what you have to significantly rotate some image, and there’s just not enough space for that. In such cases, you’ll inevitably have to deal with the blank canvas at the sides. Well, at least we can fix it fast. If the background is somewhat consistent and the model is far from the edges, you can just use the Rectangular Marquee Tool to select the white areas from both sides, and then just hit the Backspace key and Fill the selected areas with Content-Aware. Photoshop will analyze the areas around and use some appropriate contents to fill the white gaps with. And you know what, most of the time it works fine.

If this canvas extension thing occurs rather often and you need a way to speed it up, you can set the canvas extension color to something really standing out, some color that you don’t see in your images at all. Then you can select it automatically with the Select Color Range command and pick this exact color with no fuzziness, expand the selection a bit to make sure everything you need is there, and then fill with Content-Aware. But beware that sometimes this auto-filling algorithm fails, and if the same exact color you use for the extension is present in the image, it’s going to be a mess.

What I would do in this case is tell the head of photographers or whoever is responsible, that all the photographers should be taking a step or two back when shooting and make the model smaller in the frame because otherwise, it’s a hard time for the retouchers. This is one of the things easier done in the real world than during the postprocessing.

Sharpening

Customers find sharp images more attractive, and for a reason. Our eyes love details

Sharpening is a very important part of retouching, as it can make any image great or terrible, depending on the settings. As we are retouching images for online catalogues, we have to keep in mind, that in this business the overall sharpness of an image is less important than the image size in kilobytes: the bigger it is, the more time it takes for it to load on a webpage. And the only way to reduce the image size apart from making it smaller in pixels is reducing its quality. Sharpness is the main thing that gets affected when quality is reduced, so we leave all the complicated multi-stage sharpening methods for the printing business and keep it straight and simple. What does it mean?

If the original image sharpness is normal, we sharpen it once, after resizing, and we refrain from using any sharpening methods that affect color or tone. If you want your color and tone to stay intact, you should avoid big radii when applying any sharpening methods, avoiding any unpleasant post effects like halos at the same time. Small radius sharpening can be disastrous, too, if the amount is too big. The goal is to make all the images look decent, as sharpness can really increase visual attractiveness, but be careful not to overdo it.

In the following video, you might not be able to see sharpening as you would on your Photoshop screen. That's because of the video format – even in the HD quality, which is recommended when you watch videos on this website, you might not be able to perceive subtle texture changes such as those occurring when we sharpen the images. To be able to see the effect for real, repeat the same in Photoshop.

Some contents or functionalities here are not available due to your cookie preferences!

This happens because the functionality/content marked as “Google Youtube” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.

There are many sharpening algorithms readily available in Photoshop, it’s up to you which one to choose. There’s no such thing as a single best method. But I’ll show you how I sharpen all the images, and if you like it, you can just do the same. I usually record the following steps as an action and run it right after the resizing action. To do it automatically, I just make the resizing action run the sharpening action in the end. It consists of just three steps. The first one is the “Select 100% menu item”, which can be recorded by clicking on the Insert Menu Item command from the Actions main menu. Then I'll go to the View menu and click on 100% command. This will make the image scale to 100% before sharpening. You can't record this by just using the shortcut while in the recording mode, you have to use the Insert Menu Item command.

The second step is the Smart Sharpen Filter, accessible from the Filter menu, Sharpen submenu. Now press on the gear icon and check the “Use legacy” checkbox, because the newer algorithm doesn’t work as intended. Then pick a really small radius. I am used to working with images less or equal to 1500 pixels in height, and for images like that 0.2 is a good radius. The last thing you need to do is set the Amount to 500% and press Enter. Yes, it’s really a lot of sharpening, it’s too much, but don’t worry. Right after the sharpening takes effect, press Ctrl-Shift-F to access the Fade panel. Set the Fade amount to something about 30 and change the blending mode to Luminosity. It will deal with any saturation shifts due to sharpening. As my experience tells me, model images should be faded to 30% and object images can be faded to 50%, and you can just fade pretty much everything to 40% and be fine with that. More than 50 will most probably oversharpen any image

Remember, that to see how the sharpening affects your image, you have to look at it at exactly 100% scale, not more, not less. Otherwise, you’re just wasting time. Keep in mind that you shouldn’t sharpen images twice with the same algorithm, so if it feels like it’s not enough, access the Fade panel right after the sharpening takes effect and modify the amount of fading. The more percent, the sharper the image.

There is one more way of sharpening catalogue images which I’d like to show, but it should only be applied to object images like shoes, bags and so on. This time I’ll be using the Unsharp Mask algorithm, which is found in the Filter menu, Sharpen submenu. I’ll set the Amount to 20, and the Radius to 10. The result will make the objects stand out a little bit and emphasize the texture. It makes leather look especially great. But, of course, there are disadvantages as well: it does shift the tone a little bit, so watch out. As for model images, they don’t like this kind of treatment, but it’s just my opinion, perhaps you find it appropriate. So if you want to use this method, do it after the Smart sharpening and fading.

Open

As for the amount of sharpening, there's no single standard. Do a bit of experimentation to find the best combination that enhances your images, and then just stick to these values, as it’s good for consistency. It doesn’t really matter how you sharpen them, just make sure they all the images are sharpened and sharpened in the same way. Remember, that it’s better not to sharpen at all than to produce over-sharpened images, and that there’s no way to fix them afterward.

Moire after sharpening

We've dealt with moire already, as there's a huge separate chapter dedicated to the matter (LINK). But there's one more kind of moire that might bother you – the one that appears on the image after you've sharpened it. Don't worry: what's done to the image by our own hands can be undone. Just make sure you remember about this issue so that you don't get surprised when that happens.

Some contents or functionalities here are not available due to your cookie preferences!

This happens because the functionality/content marked as “Google Youtube” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.

Sometimes you'll get moire on the image after you've sharpened it. It happens with fine patterns and it might be a problem, but it can be resolved very easily by fading the final sharpening of the image in the Darken mode instead of Luminosity. To do so, just press Ctrl-F right after the resizing action has finished its work, and you’ll access the Fade panel, where the Fade effect is already set to something like 30 or 50 depending on your needs and it works in the Luminosity mode. Change the mode to Darken and you’ll immediately see the effect. If it worked, just press OK and you’re good. Sometimes it’s also reasonable to twiddle with the main effect slider and see which percentage value is the best.

Open

But in some cases, it won’t work at all. If it doesn’t, the next thing you should is going back in history and trying the stage resizing action that we recorded in the chapter dedicated to interpolation moire. Photoshop interpolation algorithms can do a better job than that, we just need to lend a helping hand. Remember, that every time we resize an image, it loses sharpness. So as we don’t want the image to be very blurry, I’d be very careful with the stage resizing action if I were you. Every single time I use it, I immediately run the final size action and check if the result is good or not. If not, I use the stage size action and check again. Most of the time I get good results with two or three runs, but it depends on the original size, of course. Be sure not to confuse moire ripples that appear after resizing to other types of moire, as you won’t be able to suppress them with stage resize or sharpening fade.

You should also remember that this kind of moire is connected with resizing, and it might appear if you view an image with some random scale numbers. The only way to evaluate the image, to see if it has moire or not is to view it at 100% scale. Any other values might be misleading.

Next chapter: Image Alignment

Top

On this website, we use first or third-party tools that store small files (cookies) on your device. Cookies are normally used to allow the site to run properly (technical cookies), to generate navigation usage reports (statistics cookies) and to suitable advertise our services/products (profiling cookies). We can directly use technical cookies, but you have the right to choose whether or not to enable statistical and profiling cookies. Enabling these cookies, you help us to offer you a better experience. If you choose to disable cookies, you won’t be able to watch the embedded videos. Cookie policy