How to freely transform while keeping the top distance unchanged?

[w30290083o9nn]

I want to use a JSX script to freely transform the image. My image has two horizontal reference lines: one at the top of the head and one at the chin. When I perform a free transform to the chin position, the distance between the top of the head and the top reference line creates a gap. How can I keep the top of the head aligned with the top reference line while also ensuring that the chin stays aligned with the bottom reference line?

[c.pfaffenbichler]

Do sou want the Script to pause and manually transform the Layer or …? 

[w30290083o9nn]

I hope to automatically adjust the layer through the script, not pause for manual transformation. Thank you

[c.pfaffenbichler]

How would you imagine that to work? 

Are you willing to mark the top of the head and the chin manually? 

[w30290083o9nn]

I have already completed manually marking the top of the head and the chin, just left with the free transform.

[c.pfaffenbichler]

You mean the guides? 

But how did mark the top of the head and the chin of the placed image itself? 

[w30290083o9nn]

I manually drew reference lines for each image one by one.

[c.pfaffenbichler]

I manually drew reference lines for each image one by one.

By @w30290083o9nn

But where are the markings in the image you provided? 

The guides appear to mark the target positions, not the source positions. 

[w30290083o9nn]

In a PSD file, the reference line at the top is the source position, and the line at the bottom is the target position.

[c.pfaffenbichler]

But how did you define the source position – the actual chin position in the actual image? 

[w30290083o9nn]

I need to scale from the top of my head to my chin to a specific height, which means placing everyone's face in the image at a uniform coordinate.

[c.pfaffenbichler]

Let me try to be clearer: 

If facial recognition does not suffice then obviously YOU need to define the source positions A1 and A2. 

Which can then be automatically transformed to the target positions B1 and B2. 

That A1 and B1 are identical is pretty much irrelevant for the automation itself. 

[w30290083o9nn]

Okay, thank you for your answer, thanks!

[c.pfaffenbichler]

A while back you wrote »I manually drew reference lines for each image one by one.«

Did you mean only the target positions by that? 

Or have you already defined the source positions? 

Defining source and target positions via guides in the same image seems problematic, so one could use a Selection, a Path, … to define the source positions. 

[w30290083o9nn]

That was to define the target positions. The target positions are the same for each image, but the height of the face and the position of the chin vary in each image.Thank you for your patient response. I will try other methods to solve this problem. Thank you again!

[c.pfaffenbichler]

Semi-automation is definitely possible, it might suffice to click on the top and bottom/chin and then run a Script. 

I wonder if one could utilize the new Object Selection Tool options (in Photoshop (Beta)) or Camera Raw Masking options to get a more precise chin-position. 

[w30290083o9nn]

Yes, semi-automation is achievable. However, obtaining an accurate chin position might be difficult. I have tried recording the Camera Raw filter as an action, but unfortunately, the recording still applies to the previous image.

[c.pfaffenbichler · Accepted solution]

@jazz-y provided a Script for face-recognition utilizing the Liquify Filter, not sure if it would be exact enough for the chin-position. 

https://community.adobe.com/t5/photoshop/face-detection-in-script/td-p/12120504

[w30290083o9nn]

I have tried this, but it couldn't achieve accuracy, so I switched to manual.

[c.pfaffenbichler]

This would raise the transform with a custom Reference Point, so alt-dragging the handles would center there. 

// 2025, use it at your own risk;
if (app.documents.length > 0) {
var originalRulerUnits = app.preferences.rulerUnits;
app.preferences.rulerUnits = Units.PIXELS;
freeTransformWithRefernce ([750,750]);
app.preferences.rulerUnits = originalRulerUnits;
};
////// free transform //////
function freeTransformWithRefernce (thePoint) {
try {
    var idtransform = stringIDToTypeID( "transform" );
    var desc6 = new ActionDescriptor();
    var idfreeTransformCenterState = stringIDToTypeID( "freeTransformCenterState" );
    var idquadCenterState = stringIDToTypeID( "quadCenterState" );
    var idQCSIndependent = stringIDToTypeID( "QCSIndependent" );
    desc6.putEnumerated( idfreeTransformCenterState, idquadCenterState, idQCSIndependent );
    var idposition = stringIDToTypeID( "position" );
        var desc7 = new ActionDescriptor();
        var idhorizontal = stringIDToTypeID( "horizontal" );
        var idpixelsUnit = stringIDToTypeID( "pixelsUnit" );
        desc7.putUnitDouble( idhorizontal, idpixelsUnit, thePoint[0] );
        var idvertical = stringIDToTypeID( "vertical" );
        var idpixelsUnit = stringIDToTypeID( "pixelsUnit" );
        desc7.putUnitDouble( idvertical, idpixelsUnit, thePoint[1] );
    var idpaint = stringIDToTypeID( "paint" );
    desc6.putObject( idposition, idpaint, desc7 );
    var idoffset = stringIDToTypeID( "offset" );
        var desc8 = new ActionDescriptor();
        var idhorizontal = stringIDToTypeID( "horizontal" );
        var idpixelsUnit = stringIDToTypeID( "pixelsUnit" );
        desc8.putUnitDouble( idhorizontal, idpixelsUnit, -0.000000 );
        var idvertical = stringIDToTypeID( "vertical" );
        var idpixelsUnit = stringIDToTypeID( "pixelsUnit" );
        desc8.putUnitDouble( idvertical, idpixelsUnit, -0.000000 );
    var idoffset = stringIDToTypeID( "offset" );
    desc6.putObject( idoffset, idoffset, desc8 );
    var idwidth = stringIDToTypeID( "width" );
    var idpercentUnit = stringIDToTypeID( "percentUnit" );
    desc6.putUnitDouble( idwidth, idpercentUnit, 100 );
    var idheight = stringIDToTypeID( "height" );
    var idpercentUnit = stringIDToTypeID( "percentUnit" );
    desc6.putUnitDouble( idheight, idpercentUnit, 100 );
    var idreplaceLayer = stringIDToTypeID( "replaceLayer" );
        var desc9 = new ActionDescriptor();
        var idfrom = stringIDToTypeID( "from" );
            var ref2 = new ActionReference();
            var idlayer = stringIDToTypeID( "layer" );
            ref2.putIdentifier( idlayer, 3 );
        desc9.putReference( idfrom, ref2 );
        var idto = stringIDToTypeID( "to" );
            var ref3 = new ActionReference();
            var idlayer = stringIDToTypeID( "layer" );
            ref3.putIdentifier( idlayer, 3 );
        desc9.putReference( idto, ref3 );
    var idtransform = stringIDToTypeID( "transform" );
    desc6.putObject( idreplaceLayer, idtransform, desc9 );
executeAction( idtransform, desc6, DialogModes.ALL );
} catch (e) {}
};

[w30290083o9nn]

First of all, thank you for your answer. Before this, I did try my own code, but there was a gap at the top position.

#target photoshop

// Get the active document and its guides
var doc = app.activeDocument;
var guides = doc.guides;

// Initialize arrays for vertical and horizontal guides
var verticalGuides = [];
var horizontalGuides = [];

// Loop through guides and separate them into vertical and horizontal
for (var i = 0; i < guides.length; i++) {
    var guide = guides[i];
    if (guide.direction == Direction.VERTICAL) {
        verticalGuides.push(guide.coordinate);
    } else {
        horizontalGuides.push(guide.coordinate);
    }
}

// Determine the left and right boundaries
var left = Math.min.apply(null, verticalGuides);
var right = Math.max.apply(null, verticalGuides);
var bottom = Math.max.apply(null, horizontalGuides);

// Calculate the selection height dynamically
var selectionHeight = doc.height - bottom; // Adaptive height
var selectionRegion = [
    [left, bottom],
    [right, bottom],
    [right, bottom + selectionHeight],
    [left, bottom + selectionHeight]
];

// Select the region
doc.selection.select(selectionRegion);

// Calculate JS_gaodu (height difference)
var X_gaodu = selectionHeight.toString().replace(' px', ''); // Get the selection height and remove 'px'
var JS_gaodu = 259 - parseFloat(X_gaodu); // Calculate the difference

doc.selection.deselect(); // Deselect the selection

// Get the layer's boundary information
var layer = doc.activeLayer;
var layerBounds = layer.bounds;
var layerWidth = layerBounds[2].value - layerBounds[0].value;
var layerHeight = layerBounds[3].value - layerBounds[1].value;

// Calculate the target height and the scale percentage
var targetHeight = layerHeight + JS_gaodu; // Calculate the target height
var scale = targetHeight / layerHeight * 100; // Calculate the scale percentage

// Resize the layer, maintaining the aspect ratio
layer.resize(scale, scale, AnchorPosition.TOPCENTER);

[w30290083o9nn]

Can this only be adjusted manually for now? Thank you again!