Bug 426349

I'm not sure about Component / Version to set (4.4.0 and 5.0.0 are not in version list)

Grum999

It's technically a regression caused by making the Fill Layer multi-threaded, but I think just adding waitForDone() in the FillLayer.cpp should solve this. I haven't checked it yet though.

(If you want @Grum check it, please see if adding waitForDone() instead of sleep() would make it work).

d52e55f6cc0567ac75c8

In script example, I've replaced 
> sleep(sv)

by 
> newDocument.waitForDone()


There's no change in results, it's still incorrect


Grum999

Created attachment 131537 [details]
Improved Python script showing all three cases

I attached an improved script that shows all three cases:
- using doc.waitForDone()
- using 1 ms sleep (not enough for Krita to finish generating)
- using 125 ms sleep (enough)
First two cases give bad results.

Steps to reproduce:
1. Open Krita.
2. Go to Tools -> Scripts -> Scripter.
3. Paste the script and execute.

I've made some other tests.

On my side it seems that to get expected results, call to sleepOrWait() is needed only after:
- a call to layer.setGenerator()
- a call to layer.setBlendingMode()

When other call to sleepOrWait() function are in comment, expected result is still OK.

With this I can't affirm that problem doesn't occurs for mergeDown(), addChildNode() or other layer/document manipulation, but this tends to confirm that problem, as suspected bu @Tiar, is relative to fill layer.


Grum999

Git commit 74dbabc3636fd5a1c18967d9fed105fb2ddaf7b2 by Dmitry Kazakov.
Committed on 16/09/2020 at 11:37.
Pushed by dkazakov into branch 'master'.

Fix Python API for Node::addChildNode()

The function should use the command, run it in a stroke and wait
for it to finish. That ensures that 1) updates are issued; 2) undo
item is created; 3) no threading issues arise.

M  +12   -4    libs/libkis/Node.cpp

https://invent.kde.org/graphics/krita/commit/74dbabc3636fd5a1c18967d9fed105fb2ddaf7b2

Git commit 57dbf8a946d022a86855b95f278f1013f29d3148 by Dmitry Kazakov.
Committed on 16/09/2020 at 11:37.
Pushed by dkazakov into branch 'master'.

Fix Python API for Node::setBlendingMode()

The function should use the command, run it in a stroke and wait
for it to finish. That ensures that 1) updates are issued; 2) undo
item is created; 3) no threading issues arise.

M  +9    -1    libs/libkis/Node.cpp

https://invent.kde.org/graphics/krita/commit/57dbf8a946d022a86855b95f278f1013f29d3148

Git commit 623c5114e009485391be9fbe943a1d9cd4da6bd3 by Dmitry Kazakov.
Committed on 16/09/2020 at 11:37.
Pushed by dkazakov into branch 'master'.

Fix KisGeneratorLayer support KisDelayedUpdateNodeInterface

It is needed to make sure the layer is updated when some synchronous
non-gui provoked actions are requested. E.g. saving or changing via
python API.

The path also fixes FillLayer's Pythong API to be synchronous using
this fix.

M  +15   -0    libs/image/generator/kis_generator_layer.cpp
M  +7    -1    libs/image/generator/kis_generator_layer.h
M  +6    -0    libs/libkis/FillLayer.cpp

https://invent.kde.org/graphics/krita/commit/623c5114e009485391be9fbe943a1d9cd4da6bd3

After my patches the following script works fine:

===========================================================

from krita import *
from PyQt5.Qt import *
from PyQt5 import QtCore
from PyQt5.QtCore import (
        QByteArray,
        QPoint
    )
from PyQt5.QtGui import (
        QColor,
        QImage,
        QPixmap
    )


def buildQImage(color, w, h):
    """Generate a QImage to use for example"""
    img=QImage(w, h, QImage.Format_ARGB32_Premultiplied)
    img.fill(Qt.transparent)
    pxmTgt = QPixmap.fromImage(img)
    
    gradientb = QRadialGradient(QPointF(w/3, h/4), 2*w/3)
    gradientb.setColorAt(0, color)    
    gradientb.setColorAt(1, Qt.black)
    
    gradientf = QRadialGradient(QPointF(w/3, h/4), 2*w/3)
    gradientf.setColorAt(0, Qt.white)
    gradientf.setColorAt(0.15, Qt.transparent)
    gradientf.setColorAt(1, Qt.transparent)

    canvas = QPainter()
    canvas.begin(pxmTgt)
    canvas.setRenderHint(QPainter.Antialiasing)

    canvas.setPen(Qt.NoPen)

    canvas.setBrush(gradientb)
    canvas.drawEllipse(QRect(20,20,w-40, h-40));

    canvas.setBrush(gradientf)
    canvas.drawEllipse(QRect(20,20,w-40, h-40));

    canvas.end()
    return pxmTgt.toImage()

def setLayerFromQImage(layerNode, image):
    """Set QImage as layer content"""
    position = QPoint(0, 0)
    ptr = image.bits()
    ptr.setsize(image.byteCount())

    layerNode.setPixelData(QByteArray(ptr.asstring()), position.x(), position.y(), image.width(), image.height())

def sleep(value):
    """Do a sleep of `value` milliseconds"""
    loop = QEventLoop()
    QTimer.singleShot(value, loop.quit)
    loop.exec()
    
dWidth=500
dHeight=500

# Cyan, Magenta, Yellow: colors used to generate layers
colors=[QColor(0,255,255), QColor(255,0,255), QColor(255,255,0)]
sleepValues=[0, 125]

image=buildQImage(QColor(255,0,0), dWidth, dHeight)

newDocument = Krita.instance().createDocument(dWidth, dHeight, "Test", "RGBA", "U8", "", 300.0)
Krita.instance().activeWindow().addView(newDocument)


parentGroupLayer = newDocument.createGroupLayer('Group layer')
newDocument.rootNode().addChildNode(parentGroupLayer, None)

for i in range(len(colors)):
    # loop over colors and apply basics actions like plugin Newspaper (and Channel2Layers) does
    newPLayer = newDocument.createNode(f"PaintLayer{i}", 'paintlayer')
    setLayerFromQImage(newPLayer, image)
    parentGroupLayer.addChildNode(newPLayer, None)
    
    infoObject = InfoObject();
    infoObject.setProperty("color", colors[i])
    selection = Selection();
    selection.select(0, 0, dWidth, dHeight, 255)

    newFLayer = newDocument.createFillLayer(f"Color{i}", "color", infoObject, selection)
    parentGroupLayer.addChildNode(newFLayer, newPLayer)

    # mandatory as when provided to createFillLayer(), infoObject is not applied
    # must also be applied after node has been added to parent...
    newFLayer.setGenerator("color", infoObject)
    newFLayer.setBlendingMode('add')
                
    newLayer = newFLayer.mergeDown()
            
    # returned layer from merge can't be used, so get the last one
    currentProcessedLayer = parentGroupLayer.childNodes()[-1]
    currentProcessedLayer.setBlendingMode('multiply')

===========================================================

Though other methods of Python API still have the same problems

Git commit 39bf729d88e70168e1e9ce58cac08fcb5f11f6f3 by Dmitry Kazakov.
Committed on 16/09/2020 at 12:19.
Pushed by dkazakov into branch 'krita/4.3'.

Fix KisGeneratorLayer support KisDelayedUpdateNodeInterface

It is needed to make sure the layer is updated when some synchronous
non-gui provoked actions are requested. E.g. saving or changing via
python API.

The path also fixes FillLayer's Pythong API to be synchronous using
this fix.

# Conflicts:
#	libs/libkis/FillLayer.cpp

M  +15   -0    libs/image/generator/kis_generator_layer.cpp
M  +7    -1    libs/image/generator/kis_generator_layer.h
M  +7    -0    libs/libkis/FillLayer.cpp

https://invent.kde.org/graphics/krita/commit/39bf729d88e70168e1e9ce58cac08fcb5f11f6f3

Git commit 7be91fc9922182bf02a3fc95feb9462ae8948f72 by Dmitry Kazakov.
Committed on 16/09/2020 at 12:20.
Pushed by dkazakov into branch 'krita/4.3'.

Fix Python API for Node::setBlendingMode()

The function should use the command, run it in a stroke and wait
for it to finish. That ensures that 1) updates are issued; 2) undo
item is created; 3) no threading issues arise.

M  +9    -1    libs/libkis/Node.cpp

https://invent.kde.org/graphics/krita/commit/7be91fc9922182bf02a3fc95feb9462ae8948f72

Git commit 7013fef0227028316c8788e828e28a66f026c08d by Dmitry Kazakov.
Committed on 16/09/2020 at 12:20.
Pushed by dkazakov into branch 'krita/4.3'.

Fix Python API for Node::addChildNode()

The function should use the command, run it in a stroke and wait
for it to finish. That ensures that 1) updates are issued; 2) undo
item is created; 3) no threading issues arise.

M  +12   -4    libs/libkis/Node.cpp

https://invent.kde.org/graphics/krita/commit/7013fef0227028316c8788e828e28a66f026c08d

Now it is just a bug, not a regression

Reassigning back to Amyspark :)