UDIM are not properly imported if spaces are present in the filename #102552

New Issue

Alexandre Labedade · 2022-11-16T10:44:02+01:00

2022-11-16 10:44:02 +01:00

System Information
Operating system: Windows 10
Graphics card: NVidia RTX 2070

Blender Version
Broken: version: 3.3.0, branch: master, commit date: 2022-09-06 15:39, type: release
Worked: version: 3.2.2, branch: master, commit date: 2022-08-02 18:15, type: release

Short description of error
If a space character is present in the filename, the UDIM tiles will not be detected, and only the first image will be imported.

Exact steps for others to reproduce the error

create some udim images with the names "udim 1001.png","udim 1002.png","udim 1003.png", etc... on your hard drive
open blender in the image editor
click the open image button, and select all the udim images to load
in Blender 3.2.2 it is properly recognized as UDIM tiles, since 3.3.0 is it loaded as an image sequence, with no possibility to properly load the udim.

If the spacing characters are replaced by a _ in the filenames, then everything is fine in 3.3.0.

**System Information** Operating system: Windows 10 Graphics card: NVidia RTX 2070 **Blender Version** Broken: version: 3.3.0, branch: master, commit date: 2022-09-06 15:39, type: release Worked: version: 3.2.2, branch: master, commit date: 2022-08-02 18:15, type: release **Short description of error** If a space character is present in the filename, the UDIM tiles will not be detected, and only the first image will be imported. **Exact steps for others to reproduce the error** - create some udim images with the names "udim 1001.png","udim 1002.png","udim 1003.png", etc... on your hard drive - open blender in the image editor - click the open image button, and select all the udim images to load - in Blender 3.2.2 it is properly recognized as UDIM tiles, since 3.3.0 is it loaded as an image sequence, with no possibility to properly load the udim. If the spacing characters are replaced by a _ in the filenames, then everything is fine in 3.3.0.

Alexandre Labedade commented

2022-11-16 10:44:02 +01:00

Added subscriber: @Alesk

Jesse Yurkovich commented

2022-11-16 11:03:41 +01:00

Added subscriber: @deadpin

Jesse Yurkovich commented

2022-11-16 11:03:41 +01:00

Changed status from 'Needs Triage' to: 'Needs User Info'

Jesse Yurkovich commented

2022-11-16 11:03:41 +01:00

This was somewhat of a conscious decision due to UDIM detection picking up numbers in the filename incorrectly in many instances - so I changed it to be more conservative and only find numbers bound on both sides by some delimiter (not a space).

For 3.3+, simply change the Image source to 'UDIM Tiles' and type the exact location of <UDIM> in the filename and it should load fine - see below:

Do you think this is alright? Have you noticed if any other software (don't have to name them) automatically detects such files as UDIMs?

This was somewhat of a conscious decision due to UDIM detection picking up numbers in the filename incorrectly in many instances - so I changed it to be more conservative and only find numbers bound on both sides by some delimiter (not a space). For 3.3+, simply change the Image source to 'UDIM Tiles' and type the exact location of `<UDIM>` in the filename and it should load fine - see below: ![image.png](https://archive.blender.org/developer/F13924186/image.png) Do you think this is alright? Have you noticed if any other software (don't have to name them) _automatically_ detects such files as UDIMs?

Alexandre Labedade commented

2022-11-16 11:28:55 +01:00

In #102552#1446745, @deadpin wrote:
This was somewhat of a conscious decision due to UDIM detection picking up numbers in the filename incorrectly in many instances - so I changed it to be more conservative and only find numbers bound on both sides by some delimiter (not a space).

For 3.3+, simply change the Image source to 'UDIM Tiles' and type the exact location of <UDIM> in the filename and it should load fine - see below:
Do you think this is alright?

It's working but not very user friendly.

Have you noticed if any other software (don't have to name them) automatically detects such files as UDIMs?

I am not sure there is a perfect automated solution for this.

For instance, in modo they give the choice to the user, see here at the bottom of the page : https://learn.foundry.com/modo/901/content/help/pages/uving/udim_workflow.html

I think that a good automated solution would be to split the filenames in chunks of numerical and non-numerical characters, and then find the pattern for UDIM tiles.
Relying on separators here doesn't seems to be flexible enough.

I have no skills in C/C++, but I could provide a python mockup of this kind of function, if you are interested.

> In #102552#1446745, @deadpin wrote: > This was somewhat of a conscious decision due to UDIM detection picking up numbers in the filename incorrectly in many instances - so I changed it to be more conservative and only find numbers bound on both sides by some delimiter (not a space). > > For 3.3+, simply change the Image source to 'UDIM Tiles' and type the exact location of `<UDIM>` in the filename and it should load fine - see below: > Do you think this is alright? It's working but not very user friendly. > Have you noticed if any other software (don't have to name them) _automatically_ detects such files as UDIMs? I am not sure there is a perfect automated solution for this. For instance, in modo they give the choice to the user, see here at the bottom of the page : https://learn.foundry.com/modo/901/content/help/pages/uving/udim_workflow.html I think that a good automated solution would be to split the filenames in chunks of numerical and non-numerical characters, and then find the pattern for UDIM tiles. Relying on separators here doesn't seems to be flexible enough. I have no skills in C/C++, but I could provide a python mockup of this kind of function, if you are interested.

Jesse Yurkovich commented

2022-11-16 11:48:25 +01:00

Changed status from 'Needs User Info' to: 'Confirmed'

Jesse Yurkovich self-assigned this 2022-11-16 11:48:25 +01:00

Jesse Yurkovich commented

2022-11-16 11:48:25 +01:00

I'll claim this for now and think about it some more. The filename is already parsed to find the numerical portion but it was "too good" at finding numbers that were not udims, like dates or image resolutions that also fall inside the udim range we support (eg 1999 or 1024 etc.) Enforcing the number to also be between delimiters of _,-, or . was done to eliminate most common occurances, and if automatic detection fails, a quick fix up like above should always be possible where the user directly indicates where in the filename things should be.

Alexandre Labedade commented

2022-11-16 11:58:19 +01:00

Ok, I see.

In #102552#1446765, @deadpin wrote:
a quick fix up like above should always be possible where the user directly indicates where in the filename things should be.

In this case, it should be mentionned somewhere in the UI, to help the user.

Anyway, I'll think about it and try to find a better automated solution, and let you know if I find something interesting.

Ok, I see. > In #102552#1446765, @deadpin wrote: > a quick fix up like above should always be possible where the user directly indicates where in the filename things should be. In this case, it should be mentionned somewhere in the UI, to help the user. Anyway, I'll think about it and try to find a better automated solution, and let you know if I find something interesting.

Alexandre Labedade commented

2022-11-23 23:45:19 +01:00

Hi,

Here is prototype python code to identify the UDIM pattern. Let me know if it could be of some help, or if I should focus on some modifications.

import re
import numpy as np

# Dummy list of files
files = []
for i in range(1001,1011):
    files.append("my file on 16-11-2022 is udim tile number "+str(i)+".png")

# RegEx pattern to detect 4 digits numbers
pattern = re.compile(r"([0-9]{4})", re.IGNORECASE)

resultA = re.split(pattern, files[0])
resultB = re.split(pattern, files[1])

# will contain an array of strings
reference = None

# will contain an array of booleans
comparisonFilter = None

# loop on files names
for filename in files:
    
    if reference is None:
        reference = np.array(re.split(pattern, filename))
    else:
        splitResult = np.array(re.split(pattern, filename))
        
        comparison = np.compare_chararrays(reference, splitResult, "==", rstrip = True)
        reference = splitResult
        
        - compare the array of bools,
        - they should contain only one occurence of False at the position of the udim number
        if comparisonFilter is None:
            comparisonFilter = comparison
        else:
            # compare the current array of bool with the new one
            comparisonFilter = np.logical_and(comparisonFilter,comparison)
        
        indices = [i for i, x in enumerate(comparisonFilter) if x == False]
        if len(indices) == 1:
            element = splitResult[indices[0]]
            
            if re.match(pattern,element):
                if int(element) < 1001 or int(element) > 9999:
                    # out of the range of UDIMS
                    print("error")
                else:
                    print("ok")
            else:
                # this is not a 4 digits number
                print("error")
        else:
            # if False value is not present, or more than one, then there is a problem
            print("error")
            break 

if comparisonFilter is not None:
    indices = [i for i, x in enumerate(comparisonFilter) if x == False]
    
    if len(indices) == 1:
        reference[indices[0]] = "<UDIM>"
        udim_pattern = ''.join(reference)
        print(udim_pattern)
    else:
        print("error")

Hi, Here is prototype python code to identify the UDIM pattern. Let me know if it could be of some help, or if I should focus on some modifications. ``` import re import numpy as np # Dummy list of files files = [] for i in range(1001,1011): files.append("my file on 16-11-2022 is udim tile number "+str(i)+".png") # RegEx pattern to detect 4 digits numbers pattern = re.compile(r"([0-9]{4})", re.IGNORECASE) resultA = re.split(pattern, files[0]) resultB = re.split(pattern, files[1]) # will contain an array of strings reference = None # will contain an array of booleans comparisonFilter = None # loop on files names for filename in files: if reference is None: reference = np.array(re.split(pattern, filename)) else: splitResult = np.array(re.split(pattern, filename)) comparison = np.compare_chararrays(reference, splitResult, "==", rstrip = True) reference = splitResult - compare the array of bools, - they should contain only one occurence of False at the position of the udim number if comparisonFilter is None: comparisonFilter = comparison else: # compare the current array of bool with the new one comparisonFilter = np.logical_and(comparisonFilter,comparison) indices = [i for i, x in enumerate(comparisonFilter) if x == False] if len(indices) == 1: element = splitResult[indices[0]] if re.match(pattern,element): if int(element) < 1001 or int(element) > 9999: # out of the range of UDIMS print("error") else: print("ok") else: # this is not a 4 digits number print("error") else: # if False value is not present, or more than one, then there is a problem print("error") break if comparisonFilter is not None: indices = [i for i, x in enumerate(comparisonFilter) if x == False] if len(indices) == 1: reference[indices[0]] = "<UDIM>" udim_pattern = ''.join(reference) print(udim_pattern) else: print("error") ```

Philipp Oeser removed the

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

UDIM are not properly imported if spaces are present in the filename #102552