Loop Action to Perspective Warp and Resize

[Esther Netherlands]

Hail to all the bright minds out there! 👋🏻

I would like help reducing the time it takes to combine a bunch of photos of boxes that were not all taken from the same angle.

The current steps are:

• Open up a template
• Drag and drop the required jpgs as smart layers
• Run an action to edit the first photo smart layer and perspective crop them to make all corners 90˚
• Save and Close
• Make the image background transparent
• Resize the layer to a specific size (out of proportion if needed) eg. 600x600px
• Select the next layer
• Run the action again
• etc
• When all layers are done, arrange them in a grid.

 

Already achieved

I have created an action that does almost all steps except the resize to a specific size.

Need help with:

1. Resizing the layer to a specific size
2. Loop the action; currently, when looping with a script, I get the error "Could not complete the Play command because the action is already playing."
3. When all is done, distribute all layers in the folder "PHOTOS" on a variable grid. Sometimes, I have 4, sometimes 6, and maybe I need to extend the template in the future to have space for more.

Attached is a folder of the action, scripts, and files

Documents 

I am unsure what is or is not achievable to automate, but any help is much appreciated! 🙏🏻

[Esther Netherlands]

Number 2 is solved by Mastermind c.pfaffenbichler 🙏🏻🤩

See this post

// 2024, use it at your own risk;
if (app.documents.length > 0) {
var check = isSmartObject();
while (check == true) {
app.doAction("Action 22", "Set 1")
var desc6 = new ActionDescriptor();
var ref4 = new ActionReference();
ref4.putEnumerated( charIDToTypeID( "Lyr " ), charIDToTypeID( "Ordn" ), charIDToTypeID( "Bckw" ) );
desc6.putReference( charIDToTypeID( "null" ), ref4 );
desc6.putBoolean( charIDToTypeID( "MkVs" ), false );
executeAction( charIDToTypeID( "slct" ), desc6, DialogModes.NO );
check = isSmartObject();
};
};
////// is smart object //////
function isSmartObject (theId) {
var ref = new ActionReference();
ref.putProperty (stringIDToTypeID ("property"), stringIDToTypeID ("smartObject"));
if (theId) {ref.putIdentifier( charIDToTypeID("Lyr "), theId )}
else {ref.putEnumerated( charIDToTypeID( "Lyr " ), charIDToTypeID( "Ordn" ), charIDToTypeID( "Trgt" ))};
var layerDesc = executeActionGet(ref);
var isSmartObject = layerDesc.hasKey(stringIDToTypeID("smartObject"));
return isSmartObject
};

[c.pfaffenbichler]

To help in selecting the relevant elements you could try using the Object Selection Tool and »Mask All Objects« (then combining the relevant Layer Masks) to get a starting-off point. 

Manual amendments will likely be unavoidable, though. 

[Esther Netherlands]

Can anyone (maybe c.pfaffenbichler 🙏🏻) help with the resize step?

I can find a resize layer script, but this works with percentages. I want to resize all my smart object layers to a specific pixel width and height, disregarding proportions.

OR... the Ferrari script 🤯

IF ratio width - height is 1:1 with a margin of 10% difference allowed THEN 600x600px
ELSE width is 600 px

I do not have much to offer in terms of scripting, but my gratitude would be tremendous if you are ever in need of a designer to lend a hand, I am yours 🤗

[c.pfaffenbichler]

I am busy at the moment; @Stephen Marsh , please forgive the intrusion but do you have some spare time at current? 

[Stephen Marsh]

I am busy at the moment; @Stephen Marsh , please forgive the intrusion but do you have some spare time at current? 

By @c.pfaffenbichler

Yes, all good!

[Stephen Marsh]

Can anyone (maybe c.pfaffenbichler 🙏🏻) help with the resize step?

I can find a resize layer script, but this works with percentages. I want to resize all my smart object layers to a specific pixel width and height, disregarding proportions.

OR... the Ferrari script 🤯

IF ratio width - height is 1:1 with a margin of 10% difference allowed THEN 600x600px
ELSE width is 600 px

I do not have much to offer in terms of scripting, but my gratitude would be tremendous if you are ever in need of a designer to lend a hand, I am yours 🤗

By @Esther Netherlands

So if the edited smart object canvas was 1100 x 1000 px (or vice versa) – would you be happy with the canvas being resized and cropped from the centre to 600 x 600 px canvas size, destructively deleting content?

[Esther Netherlands]

@Stephen Marsh first of all, thank you for your time! 🙏🏻
Not the canvas, the layer itself, but yes indeed. In the top post is a zip file with examples of hypothetical images. I did not add one for the if-else, though, but basically, it would be such a box but with a person behind it. In that case I would like to keep the proportions, but make it 600px width.

[Stephen Marsh]

The Smart Object (SO) layer without transformations is sized by the canvas size of the original SO. Resize the SO, save and close – and the SO layer size is then updated.

Test the following code with care on duped images as it is destructive to the SO content (of greater concern for linked SO files than embedded):

resizeSOto600px();

function resizeSOto600px() {
    var originalRulerUnits = app.preferences.rulerUnits;
    app.preferences.rulerUnits = Units.PIXELS;
    // Check if the document width is approximately equal to the document height within a tolerance of +/- 10%
    if (conditionalDifferenceResize()) {
        app.runMenuItem(stringIDToTypeID('placedLayerEditContents'));
        app.activeDocument.resizeImage (600, undefined, undefined, ResampleMethod.BICUBIC);
        app.activeDocument.resizeCanvas(undefined, 600, AnchorPosition.MIDDLECENTER);
        app.activeDocument.close(SaveOptions.SAVECHANGES);
    } else {
        app.runMenuItem(stringIDToTypeID('placedLayerEditContents'));
        app.activeDocument.resizeImage (600, undefined, undefined, ResampleMethod.BICUBIC);
        app.activeDocument.close(SaveOptions.SAVECHANGES);
    }
    app.preferences.rulerUnits = originalRulerUnits;

    function conditionalDifferenceResize() {
        var originalRulerUnits = app.preferences.rulerUnits;
        app.preferences.rulerUnits = Units.PIXELS;
        var theWidth = app.activeDocument.width.value;
        var theHeight = app.activeDocument.height.value;
        var theDifference = Math.abs((theWidth - theHeight) / ((theWidth + theHeight) / 2)) * 100;
        app.preferences.rulerUnits = originalRulerUnits;
        if (theDifference <= 10) {
            return true;
        } else {
            return false;
        }
    }
}

P.S. If the SO canvas has transparent pixels, I might need to add a step to trim to transparency before resizing.

[Esther Netherlands]

Thanks; however, I aim not to damage the Smart Object itself.

I did test out the script, but even if I would accept that I will need to adjust/damage the SO, the end result was not as desired.

- The SO is not resized out of proportions (SO box 1  is 1 : 1.07)

- The SO in the template file it is not reset to 100% and ending up smaller.

- Due to the resize, the mask does not match anymore - This step I can move in the process though.

Here are Video's of the first part of the process, where after placing the images, I do the perspective warp and masking transparency (note: currently crudely done, but this step will be replaced when I have found a solution to remove the background in a good way. Object select etc doesn't do a good job, I am looking into remove.bg plugin options).

2nd video is of the next step, the resize step, with your script @Stephen Marsh 

[c.pfaffenbichler]

Could you please provide a layered samlpe file that illustrates the exact state (transformation, Masking, …) of the SO that you want to scale to 600px x 600px? 

[Esther Netherlands]

Sure!

It is actually in the zip file I provided in the top post, but I added it to the Google folder  I made along the way. Here is the final result I am looking for Boxes.psd 

[c.pfaffenbichler]

I would recommend another workflow. 

• Apply a four-point-Vector Mask to the untransformed Smart Object. 

• Automate the transformation from there (the SO will be converted to a SO, so that the bounds of the Vector Mask will be honored in the transformation). 

The Script I adapted works based on the shortest side, so the result may be rotated, though. 

[Esther Netherlands]

😱 Oh wow, that looks like a huge amount of work you did for me, @c.pfaffenbichler. I will dive into it right now (and it will take me a bit of time) and test it out, but I wanted to thank you in advance!!

[Esther Netherlands]

Yeah, I see what you mean with the rotation, and that is indeed an issue when a person is standing behind the box. It flips it upside down. Not only that, all is transformed into a square as well. The script is amazing and fast though... 🤯

[Esther Netherlands]

I updated the Boxes.psd with a fake example (sorry should have done this from the start 😔) of a person on the box. It would be freakin amazing if the vector mask could start with the box selection, all transformations to be done based on the box (as the true aim is to get all the boxes the same dimension, and only enlarging of the vector mask as a last step was needed.

Truly I am so grateful and do not know how to repay you @c.pfaffenbichler I am learning so much from your script and way of thinking (never considered a vector mask!) already, and secretly hope you will be able to pull of one more rabbit out of that hat of yours and get me to the finish line with this process so I can start efficiently and fast working through my photos 🙏🏻

[Esther Netherlands]

@c.pfaffenbichler I tried adjusting your method, but it was too challenging for me to overcome/solve the flipped and all photos to square.

But with all the help from you, @Stephen Marsh + some google/chatPGT I did get my flow to work 💪🏻

PS: the 600px in my initial became 400px

For those interested, these are the steps + scripts:

Apply the same action on all Smart Objects

// 20-03-2024 // Run action on all Smart Objects // Based on code by c.pfaffenbichler
if (app.documents.length > 0) {
    var check = isSmartObject();
    while (check == true) {
    app.doAction("Action Name", "Action Set Name")
    var desc6 = new ActionDescriptor();
    var ref4 = new ActionReference();
    ref4.putEnumerated( charIDToTypeID( "Lyr " ), charIDToTypeID( "Ordn" ), charIDToTypeID( "Bckw" ) );
    desc6.putReference( charIDToTypeID( "null" ), ref4 );
    desc6.putBoolean( charIDToTypeID( "MkVs" ), false );
    executeAction( charIDToTypeID( "slct" ), desc6, DialogModes.NO );
    check = isSmartObject();
    };
    };
    function isSmartObject (theId) {
    var ref = new ActionReference();
    ref.putProperty (stringIDToTypeID ("property"), stringIDToTypeID ("smartObject"));
    if (theId) {ref.putIdentifier( charIDToTypeID("Lyr "), theId )}
    else {ref.putEnumerated( charIDToTypeID( "Lyr " ), charIDToTypeID( "Ordn" ), charIDToTypeID( "Trgt" ))};
    var layerDesc = executeActionGet(ref);
    var isSmartObject = layerDesc.hasKey(stringIDToTypeID("smartObject"));
    return isSmartObject
    };

Action containing 

1. Edit contents
2. Perspective Crop
3. Script Resize based on the ratio

// 20-03-2024 // Resize based on ratio // Based on code by Stephen_A_Marsh 
var originalRulerUnits = app.preferences.rulerUnits;
app.preferences.rulerUnits = Units.PIXELS;

if (conditionalDifferenceResize()) {
    app.activeDocument.resizeImage (800, 800, undefined, ResampleMethod.BICUBIC);
    app.activeDocument.resizeCanvas(800, 800, AnchorPosition.MIDDLECENTER);
} else {
    app.activeDocument.resizeImage (800, undefined, undefined, ResampleMethod.BICUBIC);
    app.activeDocument.resizeCanvas(800, undefined, AnchorPosition.MIDDLECENTER);
}
app.preferences.rulerUnits = originalRulerUnits;

function conditionalDifferenceResize() {
    var theWidth = app.activeDocument.width.value;
    var theHeight = app.activeDocument.height.value;
    var theDifference = Math.abs(theWidth / theHeight);
    if (theDifference <= 1.1 && theDifference >= 0.9) {
        return true;
    } else {
        return false;
    }
}

4. Save and Close

5. Script to Resize 50% > Align Top Left Canvas > Move into position

// 20-03-2024 // Scale and move Smart Object into position
if (app && app.documents.length > 0) {
    var doc = app.activeDocument;
    var originalRulerUnits = app.preferences.rulerUnits;
    app.preferences.rulerUnits = Units.PIXELS;
    
    ////// Transform Layer 50% //////
    runMenuItem(stringIDToTypeID('placedLayerResetTransforms'));
    doc.activeLayer.resize(50, 50)

    ////// Align Layer Top Left of Selection //////
    doc.selection.selectAll();
    var deltaX = doc.selection.bounds[0] - doc.activeLayer.bounds[0];
    var deltaY = doc.selection.bounds[1] - doc.activeLayer.bounds[1];
    doc.activeLayer.translate(deltaX, deltaY);

    ////// Move to Coordinates //////
    var docWidth = doc.width;
    var docHeight = doc.height;
    var offsetX = 2900;
    var offsetY = 2140;
    var layerBounds = doc.activeLayer.bounds;
    var layerWidth = layerBounds[2] - layerBounds[0]; // width of the layer
    var layerHeight = layerBounds[3] - layerBounds[1]; // height of the layer
    var newX = layerBounds[0] + offsetX - layerWidth; // new X coordinate
    var newY = layerBounds[1] + offsetY - layerHeight; // new Y coordinate
    doc.activeLayer.translate(newX, newY);
    }

[c.pfaffenbichler]

I have an idea about how to address the rotation and the »content-beyond-the-square«-issue but I am busy at current. 

[c.pfaffenbichler]

I adapted the Script a bit; extending the Smart Object’s Canvas will maintain the distortion, I just added 2000px instead of checking for content extending further, though. edited 2024-04-03

edited

[Esther Netherlands]

Sorry @c.pfaffenbichler I had some personal things to take care of before able to get back to my project.

I just tried your code, I got errors on lines 312 and 313, but ChatGPT's fix worked:

    // by xbytor, thanks to him;
    function cTID(s) { return app.charIDToTypeID(s); }
    function sTID(s) { return app.stringIDToTyp eID(s); }
    // reset;

I have just one "issue," again, due to my not fully considering all the possibilities. It is not only the top side that potentially contains the image I want to keep. Sometimes, something hangs over the sides or bottom, too. By the looks of my tests, it cuts all sides except the top (and creates a bottom blank space?).

But I am in aweeeee, like WOW!! No words, if this could work it would reduce my time soooo much. So I am hesitant to ask 🫣, but could you take one more crack at this for me 🙏🏻?

I have updated the PSD with some (roughly and random) versions of the sides possible.

And if no, not possible, I do want to thank you for all your great work and wisdom!

[c.pfaffenbichler]

I updated the code in the previous post, please try that. 

[Esther Netherlands]

... I am speechless, I am beyond grateful @c.pfaffenbichler 🙏🏻🙏🏻🙏🏻🙏🏻🙏🏻

It works PERFECT!! 

[c.pfaffenbichler]

Good to know it is helpful. 

But it only provides semi-automation, the 4-point-Vector Mask still needs to be created manually. 

[Esther Netherlands]

That is true, but just one step for me instead of 6+ steps is a huge time reduction. My project, therefore, seems less daunting and doable now. Thanks again. You are a generous genius! 🤩

[Esther Netherlands]

One question though @c.pfaffenbichler , just for cleanliness, if I check the original smart object, it has an additional "canvas" area and guidelines in it. Is this enlarging my files? Is it something easy to avoid/remove?