Smart Cropping. Exist in Photoshop?

[krdzine13]

I searched here but did not find a relevant post on this topic. 

Does Photoshop have a way of creating "smart crops" of images? In my search I see this option as part of Adobe Experience Manager, allowing sites to load dynamical cropped and position images to fit specific viewport sizes. Figma has a plugin that does this also.

I have created different artboards and loaded an image as a Smart Object that gets placed in each artboard but there is no "smart" function doing this way. Each instance has to be scaled and positioned manually. The AEM and Figma plugin options allow setting a focal point (or edge) so the image subject remains viewable.

This would be a massive time-saving option to be able to create an image (or import one) in Photoshop and then Export size variations to fit social media platforms, print, etc., without having to manually adjust.

[melissapiccone]

I don't know of a way to do this in PS. You could create actions to speed up your workflow. 

[krdzine13]

Thanks. Confirming what I thought. I have tried the actions and it can definitely speed up the process but lacks the "smart" functionality that makes the AEM and the Figma plugin so effective. For example, I used the Figma option to create 4 size variations of each image. Total batch was 300 originals yielding 1200 files without any interaction on my part once setup. It took about 15 minutes. Running action scripts that required dialog input/feedback, I only got through maybe 15-25 originals.

It's a pretty common need and would be great if Photoshop had the capability. 

[Signfeld]

"It's a pretty common need and would be great if Photoshop had the capability."

It does have it.

Once you get into scripts, you will realize capability is "near-unlimited."

Understandably, they can't add all of this in a menu of 8000 items, although it would be cool if they made something where you could checkmark some more common ones you may need to show up somewhere.

They do have some examples and useful scripts in the installation folders.

-

The more I try to script, the more I see how PS is "just basic tools" and you make it awesome by scripting your own "tools" and procedures. Only then can you get productive (and sure, many things are possible with actions alone).

[Signfeld]

I thought some of the Edit > Paste Special options were meant to help with this, but I can't get anything to happen either. Probably have to reread instructions...

Otherwise, I'm sure you can have a scale/transform or "Fit Image to Selection" kinda script adapt a layer to a selection/mask or dimension — or add some margin if you still want to be able to shift it around under a mask.

-

This script seems to work, but shows a dialog.

You would have to edit it to get the one option you want automatically...
https://community.adobe.com/t5/photoshop-ecosystem-discussions/how-to-resize-an-image-inside-a-selec...

I've also noticed that, for example, the Fit Height option will not respect the width of your selection, which may be what someone wants or not. It's not really "Fit to Selection" then, but can still be preferred depending on what you want, esp. if your next step is to add a mask from that selection — you will then have some wiggle room to move the pic underneath.

When using the script inside an action, you could follow it up with a mask step and unlink step.

-

There may be other scripts.

[Bojan Živković]

You may want to try Image > Variables > Define > Pixel Replacement > Method Fill or Fit, experiment and see what is better for you. By the way, you can not use Smart Object layers to replace content on it, just regular pixel based or shape layer with content to be replaced.

Here are my tutorials that can help you Easy Way to Automate Photoshop with Variables | DesignEasy (dezigneasy.com) at the botom are links to more tutorials.

[jazz-y]

In theory, we can manually specify a viewport for each image (for example, by selection tool) and save it in the file's metadata. Then get information from Photoshop about each "place" event and check source file that is being embedded for the presence of the metadata we need. If it is found, then perform the transformation of the layer, taking into account the viewport and binding to the bounds of the document. Both this will give behavior similar to what the @krdzine13 writes about. Also we can use the 'select -> subject' function to get rid of manual markup (however, the quality of the selection is highly dependent on the content of the image). If the @krdzine13 could post a few examples of images and more detailed information about his workflow (what sizes are used, what is inserted and where), then we could try to solve this problem with a script.