4.5 Bug processing jpg files

[LeanLeft 1]

There appears to be a bug processing jpg files.    If I load a jpg file and leave all of the settings as the defaults it seems to behave as expected.  If I change  the value for "Base" to something that is larger than the value for "Height" it appears to get really confused.   This works fine in Cura 4.4.1.   

[maartenw 6]

I import a jpg picture in Cura 4.6.2, but the base and height are interpreted totally wrong. In Version 4.3.0 it was OK - as it was for many years in former versions! Something must have changed between 4.4.1 and 4.5 apparently. Please, repair this bug!

[ahoeben 2,012]

10 hours ago, maartenw said:

the base and height are interpreted totally wrong

Could you please show with one or more screenshots, and explain what is wrong about it?

[maartenw 6]

I imported the same picture in Cura_4.3.0 and in Cura_4.6.2.

I see now that a field has been added to the import window: Color model.

 

The import window for Cura 4.3.0 is:

What results in:

 

The import window for Cura 4.6.2 is:

 

With the result:

( Cura 4.6.2 shows an error importing the same picture. )

Edited August 7, 2020 by maartenw

[LeanLeft 1]

Frustrating, I reported this bug back in March and supplied sample files and screen shots showing the good 4.4.1 version and the bad 4.5 version.  All those attachments seem to have disappeared.   

    

[ahoeben 2,012]

On 8/7/2020 at 2:51 PM, maartenw said:

I imported the same picture in Cura_4.3.0 and in Cura_4.6.2.

The perspective on the models is so different that I still can't see what the difference is.

 

On 8/7/2020 at 2:51 PM, maartenw said:

I see now that a field has been added to the import window: Color model.

Does changing the color model make any difference?

 

[maartenw 6]

Thanks for the reply, ahoeben.

You cab see the difference on the dimension indications: In the case of 4.3.0 the height is 10.1 (acceptable), for 4.6.2 the height is 2.6 (mm). The perspectives are the same for the pictures shown, but I had to include the error message, so the total picture is different.

I imported the same picture again, now for the Color Model 'Doorschijnend' (Transparant?), leaving the new field to the default 50. Now the dimension is the same as for Cura 4.3.0. The import again results in the error message. After empting the platform and reimporting, the result is identical, bit no error message appears. Closing Cura, restarting and then reimport shows the same result with - again - the error message.

 

Import using the Color Model as 'Doorschijnend'.

 

Give the result:

 

I hope this helps tackling the problem.

 

-> Where can I find a clarification on the new fields 'Color Model', and 'Doorschijnendheid' (Transparancy)?

Edited August 9, 2020 by maartenw

[bellzw 0]

I have been studying the behavior of the Translucency (that's the word in my English installation of Cura 4.8.0) for making lithophanes, my ultimate goal.  The error related to the model not being manifold is, I think, unrelated to the Color Model.  It's a statement about holes or other irregularities that were found when model was sliced.  There is an article about this at Cura support, article 360014055959-Why-am-I-getting-Your-model-is-not-manifold-popups-in-Ultimaker-Cura-.  When making lithophanes, I have not had a problem printing the object, despite seeing these errors.

 

I am a little confused by the screen shots since the importing of the jpg should have resulted in a model that was 12 mm (11+1) because Height, according to the screen tip, is height above the base.  Both 10.1 and 2.6 seem wrong.  However,  I won't comment further on this because I may have misinterpreted the previous comments.

 

To study how Cura 4.8.0 imports a jpg, I created an 400x300 grey-scale image with Paint having nothing but 8 square columns with shades of grey from 128 to 1, stepping down by a factor of 1/2.  The columns were on a white (pixel value 255) background.  I set the top and bottom layers to 0, layer height to 0.1, first layer thickness to 0.1 also.  I imported with Height = 25.5, Base=0, Width=120, Depth=90 using Linear and Transparency modes and Darker is higher and Lighter is higher.  I found that Linear worked correctly and gave me objects that were 25.5 mm in height, exactly what I should have gotten according to the Base and Height I set and the pixel values in the jpg.

 

Translucency darker is higher was successful, however it is not clear that the slicing is correct.  I set 1 mm transmittance = 50% and the resulting slicing had 176 layers to the top of the tallest tower, a factor of 2.2 times what should have been given the layer height was 0.1 mm.  The reason I expected an 8 mm tall model was because if the light loss is 50% in 1 mm, then there needs to be 8 mm of material to attenuate 255 to 1 (2^(-8)).  Cura did correctly scale the towers according to the logarithm (the steps in the image below are equal), but seems to have an error that makes it think the layer height is smaller than the user sets.  This factor of 2.2 does not seem to depend on the user-supplied layer height parameter.

 

Translucency fails when it is used in lighter is higher mode.  The image below shows that all the holes either go to the bottom or stop only 1 or 2 layers from the bottom.  The picture should have looked like the one above, just upside down.  This means you can't make a negative lithophane.