I have great difficulty with Perspective correction

[pledden]

I locate the bounding box corners on the corners of my distorted photo and double click inside the borders and the photo reopens with another, smaller bounding box with some of the photo cut off. What am I doing wrong?

[LeviFiction]

Perspective warp correction works like this. You find an area that's meant to be square on with the camera view, and then you align the bounding box corners with that. When you apply that correction it not everything will be perfectly square there will be areas of the canvas that show through at the edges, so if you have "Crop Image" checked it'll attempt to auto-crop to get rid of those areas. If you don't have "Crop Image" selected you'll see the background showing through. So you don't want to put it at the corners of your image you want the bounding box to reflect the perspective of the image.

Fun note, you can control the background color that shows by selecting the background color in the materials palette before applying the perspective correction.

[pledden]

Thank you for your reply. I have attached the photo I am trying to straighten. Maybe I am asking too much of the program to rectify the distortion, to start with the jigsaw is on a stand at 45 degrees and I am using a 24mm lens to get it all in. maybe I am asking the impossible?

[LeviFiction]

I'm clearly not understanding the problem. Put the nodes directly at the corners of the puzzle making sure the lines of the bounding box perfectly match the angles of the puzzle. Yes it will auto crop unless you uncheck "Crop Image" on the tool options palette. So if you don't want it to autocrop uncheck that checkbox.

Here is the result of my applying perspective warp correction on the image.

I further cropped it on my own with the crop tool to get rid of the green border.

[Jean-Luc]

Maybe I am asking too much of the program to rectify the distortion,

You need to place the corners of the straight rectangle exactly on the corners of the puzzle before applying.

Edit: Oops, didn't see Levifiction's message...

[JoeB]

Like @LeviFiction I'm not sure I understand the issue. Here's how I approached it, with 2 images attached.

First, promoted your image to a regular raster layer then increased the canvas size with the image centered to make it easier to align the perspective correction line on the left side of the image.

With the Auto Crop UNchecked, I aligned the Perspective Correction tool and I applied the correction.

Then I used the Selection rectangle tool around the image to isolate the image from the surrounding green borders then cropped to selection. The result is my first image. You'll note that this leaves a small section in the bottom left corner that is showing transparency because, of course, whatever belongs there is not in the original image. If you leave Auto Crop on the result is that it crops the image at the left, cutting off the transparent part but also cropping a portion of the actual image from the left side (note how the FESTIVAL banner near the left bottom is much closer to the edge of the image in @LeviFiction's result than it is in the original or my images).

Than I cheated a little. Again, increasing the canvas size to have room to work, I selected the Mesh Warp brush and carefully pulled the bottom left corner of the image to the left until it aligned with the rest of the upper part of the actual image.

I then again took the Selection rectangle tool, selected it around the image and cropped to selection. Doing so means that none of the image on the left is cropped out as it would be with Auto Crop enabled and the bottom corner is now filled in. Granted, it means that the image at bottom left is not the original jigsaw but, with this particular image, that wouldn't be apparent to any observer and the distortion caused in that spot by the Mesh Warp tool is likely just as unnoticeable. The result is my second image.

[pledden]

I'm afraid I have no idea how to promote to a regular raster layer or increase the canvas size or how to centre the image, maybe I am wasting your time as I am far less experienced than yourself. My apologies.

[LeviFiction]

You are not wasting anyone's time. The point of this forum is for people to ask questions and get help. We may struggle for a bit to get the right answers to you but hopefully with time and patience we'll all get there.

Let's start with the parts of JoeB's instructions that you aren't familiar with. I'll do both a simple and long explanation.
1) Promote to Regular Raster Layer - In the layer's palette right-click on the layer and select "Promote Background Layer". If this option is not available. you can skip this step.

Explanation:
In PSP there are several different layer types. The most common and basic is the "Raster" layer. This is for regular photos and image data. Rasters are made up of individual dots of color (pixels) that make up an image. When you edit a raster layer the edit is destructive, as you overwrite the color values that were there originally. And to make things even more interesting there are two types of Raster Layers. Background layers and regular full Raster layers. A Background layer does not support transparency while a Full raster layer does. Jpeg files, or RAW images, don't support transparency so when they are opened in PSP they open as Background Layers. You can tell because the layer will be named "Background" and the icon of the layer will look like a single piece of paper. When you promote the background layer you're telling PSP that you want the layer to support transparency. The layer will then be automatically renamed to Raster 1 (or if you already have raster layers in your image, the next # in the progression) and the icon will change to a sheet of paper with a transparent sheet on top of it. See the images below for examples.
Background Layer exmaple

background-layer-example.png (15.49 KiB) Viewed 791 times

Menu item to promote background layer

Final result of promoting the background layer

FullRasterLayerExample.png (15.33 KiB) Viewed 791 times

2) Increase Canvas Size and keep original image centered
- To do this go to the top menu and select Image -> Canvas Size. Then in the dialog that pops up increase the width and height by some amount. Try starting with 100 pixels each, it's hard to know exactly what you'll need so play with it. And in the "Placement" section of the dialog hit the center button to keep the original image centered.

Explanation:
Every image is placed on a "canvas" of a set size. When you open an image PSP sets the canvas size equal to the size of your image. But when you create a new image from scratch you tell PSP how big you want the canvas size to be. The command Canvas Size lets you resize the canvas without resizing the entire image. The Resize command will scale everything in the image not just the canvas, while Canvas Size only resizes the canvas making it larger or smaller. The dialog also provides the ability to determine how the original image will be placed in the new canvas. Here's an example image, you can see I've increased the Width and Height from the Original size by 100 each, and I've got the Center Placement button selected to ensure the image is centered once the command is done.

[LeviFiction]

3) Align the Nodes and apply the perspective correction.
Select the Perspective tool and put the 4 nodes at the exact corners of the puzzle (in your example image they are loosely near the corners) you want the bounding box of the tool to match your puzzle almost exactly as the puzzle shows the perspective you're attempting to correct. Make sure "crop image" is unchecked on the tool options palette before applying the correction. This prevents PSP from cropping your image smaller.

This image shows the settings you'll notice I have the checkbox "Crop Image" unchecked.

Explanation: The perspective warp uses the bounding box to tell PSP what the perspective shape is that it's attempting to correct for. In some images this might be a window, or a desk, or the angle of the road. Something is tilted and warped inside the image that you want to show head-on. You are trying to match the box as closely to the perspective it's attempting to correct for. Otherwise it'll try to correct for an incorrect perspective. So you can see in the image below how I've applied the perspective warp bounding box.

While your original example had the box generally surrounding the the puzzle inside the green border, here you can see the nodes are in the exact corners, and lines between them match the puzzle outline. Including the node that should be around the cut off corner of the puzzle. I found it's placement by making sure the line properly followed the edge of the puzzle. Leaving "Crop Image" unchecked will stop PSP from cutting off parts of the image. Giving you the flexiiblity to decide for yourself how you want the final image to look once you're done editing.

From there JoeB goes on to explain how he dealt with the corner, but I won't go into that for now.

Hopefully the explanations and images help.

[pledden]

I wanted to post this at the end but it won't let me so here goes. All your help was so invaluable and I can now correct perspective perfectly. much appreciated.

[Kathy_9]

Merged topics for you.