to make this quicker for everyone: 1. Ctrl/cmd + J to make a duplicate layer if you dont want to drag it. 2. after you get the clean background layer without the subject, go to filter, camera raw, geometry, and just have it auto adjust your perspective, auto usually works but if not it also has tools that you can use to adjust it in literally 5 seconds. everything shown in this video is a little excessive and time consuming and you can do it in native Photoshop now, you dont need beta.
In Photoshop 2024 we now can use gen expand as an option to expand an image in the Crop tool. That will keep Camera Raw's Geometry from returning empty triangles in the corners. You could select each triangle and request a Generative Fill instead of a Content-Aware fill.
Nice technique to know, but I think it's also important to know when to use it. The 2nd image really changed what you could see, and you lost a lot of the framing from the columns. The 3rd image, you're completely changing the photographer's intent, and the effect seems very intentional.
agree especially on the 3rd but keep in mind the image was only being used for educational purposes never to state that the original image was wrong or bad in any way
Camera Raw’s Geometry is a way more efficient way to do this. Perspective Warp is too cumbersome for backgrounds. Even the Lens Correction Filter can help in some light distortion cases.
Oh my gosh, the comments. This is a demo of how to use these tools, not a demo of how to create perfect images. You can only do so much in 20 minutes. Amazingly instructive as always. Looking up how to use perspective warp was on my list of todos, so very helpful!
Always very thorough and well explained. Also like the fact you aren't afraid to admit when these new innovative technologies are not perfect. Being that I am still working with CS6, I want to take a look at new features and see how well (or how not so well) some of this stuff is going to work (especially this "Generative Fill" stuff). Thanks Aaron for this very informative tutorial.
I realize there are several ways to accomplish the same task in PS; however, I found this method using the Transform and Perspective Warp tools unnecessarily complex while providing marginal results. In addition, using this method created backgrounds that are out of proportion from the original background. I followed the same steps, but went in a different direction after merging layers into the “Background CLEAN” layer. I then used the Camera Raw Filter to open the layer in ACR. Once in ACR in the Geometry panel, I used the Guided Uprights to draw two lines, one close to the left of the frame and one close to the right. Back in PS, I used the Lasso tool to draw loose selections around each transparent triangular area, and then Generative Fill. This filled in those areas quickly and very accurately. This method is quicker and easier, while creating backgrounds much closer in proportion to the original.
Wow. Such a great tutorial. I could never use the perspective warp tool before because I lost so much of the photo - but the generative fill is a complete game changer for this tool now. Thanks for the video.
any reason not to just use the guided transform tool in camera raw? Also, I love how when you guys use generative fill to remove a person, it seems to always do it clean. When I try to do it, I get some deformed AI person in their place. lol
Cool techniques, but the loss of the glasses frame over the right eye (image 1), the misalignment of the tile lines (image 2), and the thinner shadow under the right foot (image 3) are all unacceptable!
This is a fantastic demo on how to use these tools, especially now that we have Generative fill to help us along. Let's not confuse that with when or if we want to use it. Thanks, Aaron ! This was very informative.
You make the same mistake as every Photoshop teacher I've seen: adjusting the upright lines to exactly vertical is unnatural to human eyes (and brain...). Depending on the distance we normally "see" more or less falling lines - that's how our brain generates height. In your picture you should have kept some distortion because the columns are close to the subject. And moreover, the next much bigger mistake you made (like other teachers) is that by your method the columns (2nd example) get very compressed: the lady was exactly half the size of the columns - bat afte your operation she became almost the same size as the columns. Therefore, you should stretch the background also vertically!
Ironically you'd do the exact opposite if creating a composite with a cut out subject. On this example you can see the reason for the sloping background lines is because the subject was shot from down below in a crouched position camera pointing upwards. If you shot a subject in the same way and wanted to place them in a new background you would want to alter the perspective deliberately to slope inwards to make it look realistic. By straightening the background in this example you make it look like a poor composite lol 🤔
Hey there! Most likely you need to invert the selection. If you send us an email to support with a screenshot of your PSD file and layers, our team will do their best to guide you!
Hey there! Most likely you need to invert the selection. If you send us an email to support with a screenshot of your PSD file and layers, our team will do their best to guide you!
Hey Aaron, sorry, but this is your worst "tutorial" till now, really sloppy. Exactly how not to do it... To transform background (you call it "correcting the perspective") separate from the subject gives the pictures a horrible look and this ends always in a "Frankenstein perspective" where nothing connects anymore. I think that's a bad idea in general. However, I learned a lot from your videos, your work has always been great, but all this generative fill seems to get people and you too to forget how it's done right. Where is your own claim to quality and details gone? Apart from "correcting the perspective" there were some other things you just overlooked, in the first example there were some problems with the selection of the subject which could easily been fixed (glasses and hair), in the second Image there was the problem with the lines in the ground that did not match anymore and the third image has really shown that "correcting the perspective" this way really does not work. The person lost its shadow and completly hovered in the image. You really should come back to your roots - sorry man.... You better should show us how to get to a great result and not how to make pictures more worse.
