Code to run an external Python script from Krita
import subprocess
def execApplication(address):
subprocess.Popen("python " + address, shell=True)
script = r"C:\XXX\XXX.py" # input address of your python script
execApplication(script)
3 Likes
Hereās a snippet to apply a transform mask affecting a group to all paint layers inside it.
Letās say you got a group with lineart, another with colors, several layers inside. You used liquefy on a transform mask applied to one group contents to push the stylization and youāre satisfied with the results, and copied the transform to the other group too, but the performance isnāt great and Krita keeps doing that thing of updating the mask.
Youād like apply the changes to each child layer if you could without having to do it manually or merging the group. You can:
from krita import Krita
# HOW TO USE
"""
Select transform mask inside group and run script.
The mask will be copied to every paint layer inside the group,
and those layers will be flattened to apply the transform directly to them.
The original mask will then be hidden.
"""
print("\n > Preparing to transfer transform mask")
app = Krita.instance()
doc = app.activeDocument()
mask = doc.activeNode()
parentGroup = mask.parentNode()
siblings = parentGroup.childNodes()
window = app.activeWindow()
view = window.activeView()
def showWarning(msg, duration = 3000):
if view is not None:
view.showFloatingMessage(msg, Krita.instance().icon("16_light_warning"), duration, 1)
def applyMask(siblings, mask):
type = mask.type()
if type != "transformmask":
showWarning(f"Layer must be a transform mask inside a group")
print(f"Invalid layer type: '{mask.name()}' is {mask.type()}, not a mask layer")
return False
elif parentGroup.type() != "grouplayer":
showWarning(f"Parent layer '{parentGroup.name()}' is not a group")
print(f"Parent layer '{parentGroup.name()}' is not a group layer, it's a {parentGroup.type()}")
return False
# Duplicate mask or filter to apply to layers
if type == "transformmask":
dup = mask.duplicate()
if not mask.visible():
dup.setVisible(True)
for layer in siblings:
# Validate layer type
if layer.type() == "paintlayer" and layer.visible():
layer.addChildNode(dup, None)
doc.setActiveNode(layer)
Krita.instance().action('flatten_layer').trigger()
# Hide mask so it doesn't affect siblings twice
mask.setVisible(False)
print(f"> Applying mask '{mask.name()}' to all siblings in group...")
applyMask(siblings, mask)
# Force final visual update
doc.refreshProjection()
doc.waitForDone()
print("> Done applying mask to siblings.")
Word of caution: I currently donāt have the time to make it recursive, think of corner cases or do more than a cursory validation. Because itās flattening layers to apply the mask other masks and filters layers might have will also be applied.
Itāll apply the selected mask even if itās originally hidden for performance reasons.
Itās not recursive, layers inside groups inside the parent group wonāt be affected. It also only affects visible paint layers.
Also, their names are getting āMergedā appended.
4 Likes
This is a gem
2 Likes
Switch to the brush preset with the specified name
app = Krita.instance()
window = app.activeWindow()
view = window.activeView()
targetPresetName = "***"# input the name of the preset
allPresets = Application.resources("preset")
for preset in allPresets:
if preset == targetPresetName:
view.setCurrentBrushPreset(allPresets[preset])
break
3 Likes
The sample of the vector shape positioning, transform inheritance by shape.absolutetransformation() . SVG scaling setting.
from krita import *
from PyQt5.QtCore import QPointF
import re
import math
def add_rect_to_nearby(shape):
svg_data = shape.toSvg()
print("original element:")
print(svg_data)
print()
p = shape.position()
print(f'position {p.x()},{p.y()}')
absolute_transform = shape.absoluteTransformation()
abst_mat = qtransform_to_svg_transform(absolute_transform)
print(abst_mat)
# Blue rectangle: Place new svg element on near of the shape.
# this rect far x=10 y=10 from the shape
# This position will inherited from the transform of the layer and shape.
element = '<rect transform="'+abst_mat+'" x="10" y="10" width="180" height="80" fill="blue" stroke="black" stroke-width="2" />'
# inherit of transformation or keep position
pos = f' x="{p.x()}" y="{p.y()}" '
print(pos)
# inherit of transformation
element +='<text style="font-family:Arial;" '+pos+' font-size="36" fill="purple">Hello Krita!</text>'
# keep position
element +='<text style="font-family:Arial;" transform="'+abst_mat+'" x="10" y="20" font-size="36" fill="red">Hello Kiki!</text>'
return element
"""
QTransform(a, b, c, d, tx, ty)
matrix(a b c d tx ty)
a = scaleX
b = skewY
c = skewX
d = scaleY
tx = translateX
ty = translateY
transform = QTransform(1, 0, 0, 1, 100, 200)
svg_transform = qtransform_to_svg_transform(transform)
print(svg_transform) # "matrix(1 0 0 1 100 200)"
"""
def qtransform_to_svg_transform(transform: QTransform):
return f"matrix({transform.m11()} {transform.m12()} {transform.m21()} {transform.m22()} {transform.m31()} {transform.m32()})"
# -------
# main
# -------
app = Krita.instance()
doc = app.activeDocument()
view = app.activeWindow().activeView()
selected_vector_layer = view.selectedNodes()
# Set corrected scaling. 1pt = 72 dpi
wpt = doc.width()*0.72
hpt = doc.height()*0.72
svg_scale = f' width="{wpt}pt" height="{hpt}pt" viewBox="0 0 {wpt} {hpt}" '
output_shapes = []
# Get selected shapes and apply path effect to the shapes
for node in selected_vector_layer:
if node.type() == "vectorlayer":
shapes = node.shapes()
for shape in shapes:
if shape.isSelected():
# main (get an output_element string)
output_elements = add_rect_to_nearby(shape)
# When adding to Krita vector layer, you don't need to describes the DTD or XMLNS parts.
output_shapes.append("<svg "+svg_scale+">"+output_elements+"</svg>")
# Add shapes to the layer
for s in output_shapes:
node.addShapesFromSvg(s)
1 Like
Was going to make a topic asking how to get current node before/after actions are triggered, but after some back and forth with copilot apparently using lambdas containing a tuple (expression) can work. There are a couple downsides like variables being local to each expression, but you could use functions instead and time them in a similar way. Probably solves the global variable issue but then you got to edit the function names in multiple places. Iāve not tried using doc.waitForDoc() here but if this allows mixing api/actions then maybe it will work nowā¦?
from krita import *
inst = Krita.instance()
doc = inst.activeDocument()
def func():
if not doc:
print("no active document")
return
if not doc.selection():
print("no active selection")
return
steps = [
lambda: (
print("convert_selection_to_shape"),
inst.action("convert_selection_to_shape").trigger()
),
lambda: (
layer := doc.activeNode(),
print(layer),
svg := layer.toSvg(),
print(svg),
inst.action("deselect").trigger(),
),
]
delay_per_step = 100 # milliseconds
for i, step in enumerate(steps):
QTimer.singleShot(delay_per_step * (i + 1), step)
func()
Here is how I (and many others apparently) assumed it would work, but doc.activeNode() gets the previous node. This has been discussed over a bunch of topics.
from krita import *
inst = Krita.instance()
doc = inst.activeDocument()
def func():
if not doc:
print("no active document")
return
if not doc.selection():
print("no active selection")
return
inst.action("convert_selection_to_shape").trigger()
inst.action("deselect").trigger()
layer = doc.activeNode() #active layer is not the new vector layer
print(layer)
print(layer.toSvg()) # AttributeError: 'Node' object has no attribute 'toSvg'
func()
edit:
Here is a version that a list of functions. For whatever reason the prints go to the command prompt (I launched krita from there). the dictionary var is used to store values.
from krita import *
from PyQt5 import QtCore
inst = Krita.instance()
doc = inst.activeDocument()
var = {}
def func():
if not doc:
print("no active document")
return
if not doc.selection():
print("no active selection")
return
def func1():
print("convert_selection_to_shape"),
inst.action("convert_selection_to_shape").trigger()
# var['var1'] = "hello"
def func2():
print("func2")
layer = doc.activeNode()
var['svg'] = layer.toSvg()
print(var['svg'])
def func3():
print("func3")
inst.action("deselect").trigger()
pass
funcs = [func1,func2,func3]
delay_per_step = 100 # milliseconds
for i, func in enumerate(funcs):
QTimer.singleShot(delay_per_step * (i + 1), func)
#
func()
2025-06-09
Iāve been testing this for a while, now Iām using yield for each delay, itās much more easy to read, variables are all in the same scope, and loops even work. Save your work before trying this 
Not all commands will undo!
Besides the decorator function, the script I included here will select pixels from a transparency mask and apply it to a group above, then delete the layer and deselect.
from krita import *
inst = Krita.instance()
doc = inst.activeDocument()
node = doc.activeNode()
view = inst.activeWindow().activeView()
def async_runner(func):
"""Decorator to automatically handle function execution with pauses."""
import functools
@functools.wraps(func)
def wrapper():
generator = func()
def step():
try:
delay = next(generator)
QTimer.singleShot(delay, step) # Schedule next step
except StopIteration:
print('--- DELAYED EXECUTION COMPLETE ---')
step() # Start execution
return wrapper
@async_runner
def func():
layer_type = "transparencymask"
if len(view.selectedNodes()) != 1:
print(f"must select only 1 {layer_type}")
return
if node.type() != layer_type:
print(f"need to select a {layer_type}")
return
parent_type = "grouplayer"
parent = node.parentNode()
if parent.type() != parent_type:
print(f"parent node must be {parent_type}")
mask = node
yield 10
inst.action("krita_filter_invert").trigger()
yield 25
inst.action("selectopaque").trigger()
yield 25
inst.action("remove_layer").trigger()
yield 50
doc.setActiveNode(parent)
yield 25
inst.action("clear").trigger()
yield 25
inst.action("deselect").trigger()
yield 50
func()
Depending on document size and complexity of layers obviously it will not always work, maybe this can be mitigated with a function that tweaks the delay based on document sizeā¦ I would really like to be able to call other functions with their own delays so I can reuse some of these, but it seems tricky. Maybe inspect the function and count the values of yields? I think I might have tried it and it didnāt work 
1 Like
2 posts were split to a new topic: How to script left click anywhere on UI
doc.extendImage() is like crop, but it doesnāt delete pixels outside the canvas. Could be a useful option in the crop tool 
from krita import *
def msg(msg, duration=500):
Krita.instance().activeWindow().activeView().showFloatingMessage(msg, Krita.instance().icon("16_light_warning"), duration, 1)
def func():
inst = Krita.instance()
doc = inst.activeDocument()
if not doc: return
sel = doc.selection()
if not sel:
msg("Nothing is selected")
return
x = sel.x()
y = sel.y()
w = sel.width()
h = sel.height()
doc.resizeImage(x,y,w,h)
func()
1 Like
āBonsaiā scripts: 100 UNDO and 100 REDO
Sometimes in sketches or brush tests, Iād love to delete all undo stack returning to a blank canvas (EDIT: and not have the old stuff appear when I go back a few undoes in the new drawing).
I didnāt find how to do this (EDIT: with a hotkey), so, despite itās not the same thing, I use this little workaround:
# "Bonsai" script 100 UNDO
# Executes 100 undo (can be changed in "range(100)" part as desired)
from krita import *
for x in range(100):
Krita.instance().action('edit_undo').trigger()
I didnāt find also how to get how much undo actions are in the stack, so I decided to do 100 undo, but it can obviously be changed as desired.
WARNING: when you do something after that the undone stack will be lost!
Since thereās the risk to ādeleteā a good part of the work calling the script inadvertently, Iāve set also the redo.
WARNING: it must be called without doing other actions otherwise all the undone stack will be lost!
# "Bonsai" script 100 REDO
# Executes 100 redo (can be changed in "range(100)" part as desired)
from krita import *
for x in range(100):
Krita.instance().action('edit_redo').trigger()
FIRST USED IN ā Krita 5.2.9 Portable - Win 11
Advices and improvements are super welcome!
We got the Undo History Docker where you can go to every single step you made so far, even back to your empty canvas, at least if you have set up enough possible undo steps.
Michelist
Yes, but it requires a lot of clicks and scrolling, itās better to have a hotkey to do this, really quick and not distracting for me.
I didnāt find āUndo Historyā hotkey possibilities to go to the first, or other thingsā¦ am I wrong 
Why should you be wrong? It is only another possibility, nothing more, and since your question above sounded as if that and other possibilities were completely unknown to you, I made my post.
What I like with the docker, is the ability to pick out the point in history where I want to go. If scrolling here or in another app is too slow for you, simply drag the slider on the scroll bar to speed it up. This isnāt a problem for me because my mouse has a free-running flywheel that allows me to scroll at lightning speed. Of course, a mouse wheel with detents is an obstacle.
Another way to clear all undo steps at once is Kritaās Clear Button:
Michelist
AFAIK the āClearā toolbox action (āDelā hotkey) adds the āClearā action in the undo stack, it doesnāt delete the previous stack, so if you then go back with some undo the old things are all there.
EDIT: Iāve added in āBonsaiā scripts description a little explanation cause it was not very clear:
āSometimes in sketches or brush tests, Iād love to delete all undo stack returning to a blank canvas (EDIT: and not have the old stuff appear when I go back a few undoes in the new drawing).
I didnāt find how to do this (EDIT: with a hotkey), so, despite itās not the same thing, I use this little workaround:ā
Oops, you are right!
I must admit, I did not check that, because I did not have the issue that let you write your scripts, as I wrote, I can navigate extremely fast because of my mouse.
But your scripts are definitely a good way to revert 100 steps, or whatever you set up in the script, at once.
So, thank you for them, the day may come when Iām in need of them, without me knowing right now.
Michelist
1 Like
Also @Michelist the āClearā action works only in the current layer, but my āBonsaiā scripts work EVERYWHEEEEEERE ( 
) ā EDIT: everywhere in the current document.
1 Like
Go ahead and finish me off! Boo-hooā¦
Michelist
1 Like
Oh wow ! that topic is neat!
Am not the best in coding but am definitely interested in this. I was looking a way to get rid of my input-remapper because I use the freehand selection tool as an eraser and right now i make it with a macro.
Do you think, i could use a script into krita instead of the macro? Like do the krita script can use key automation or only action in the software?
And if itās only action is this reproducible?:
1. delete key # itās the clear action layer, this macro is intend to be used after the freehand selection tool
2. Ctrl+shift+A # to deselect
3. wait 0.020 sec # to let the computer breath
4. B key # to select my brush again.
is this macro scriptable?
Most of things can be done with scripting. (sometimes there are issues on execution order, giving bad time)
from krita import Krita
# cut, deselect, activate brush tool
app = Krita.instance()
app.action('edit_cut').trigger()
app.action('deselect').trigger()
app.action('KritaShape/KisToolBrush').trigger()
/AkiR
1 Like
Thatās amazing thanks.
Hello all!
Iām currently working on a set of scripts to automate simple animations.
The subject Iām working on is for an internship, and the intent is to develop a plugin that allows for a set of commands to automatically make animation frames for simple things.
In the case of the below code- this was/will be used to move a cloud across the screen, to the right.
Although, youād have to set the tool options to adjust the incremental rate of your current layerās translation. I recommend setting the units from mm to Pixels.
from krita import *
def layer_duplicate():
# Get the Krita application instance
application = Krita.instance()
# Get the active document
currentDoc = application.activeDocument()
# Get the active document and node
document = Krita.instance().activeDocument()
activeNode = document.activeNode()
#The initial value of f is the starting frame, the <= value of f is the upper bounds (the end point)
f = 0
while f <= 30:
currentDoc.setCurrentTime(f)
Krita.instance().action('movetool-move-right').trigger()
Krita.instance().action('add_duplicate_frame').trigger()
f += 1
currentDoc.refreshProjection()
layer_duplicate()
My next main goals are to make a similar setup for rotation, and, if possible, create a script to adjust the tool options (so that they may be adjusted by a user from a dialogue when itās set up through a plugin).
Much of my progress thus far has been reviewing the action-dictionary as well as these forums, and some older posts have left me questioning the current access I have to configure the program through the Scripter and whether or not thatās possible.
Per each of these goals, these are my questions for those of you who use the program on a more in-depth or technical level:
- For rotation, at least for 90 and 180%, I can manually complete the animation similarly to the code, but with the difference of needing to re-select the layer with the transform tool each time.
I can check and print the name of the activeNode in the Scripter, but this isnāt the same as clicking on it to create the free-transform box. Is there a command that Iām missing that could be added where the movetool-move-right action currently is?
- Iāve read that the tool settings are kept in a particular file so that the program may remember their settings between projects. In the most recent version of Krita, how best can I access this file, and could it also be accessed through the Scripter?
I will continue to look at the API and also futz with the manual timing of the pieces I need.
My hope is that since the topics that said that this type of configuration wasnāt an option were posted, this mightāve been approached by a developer or fellow user.
Best Wishes, and Be Well,
-RC
1 Like