Fills with "Noise" – reason for inconsistent results when printed from PDF-export?

[Lorox]

Maybe someone can shed a light on this:

I recently had some files for a signage project printed and my printer encountered mysterious problems because one of 5 signs showed a slightly but noticeably different shade of grey as its background when printed than the others.

The grey in question was a CMYK fill (consisting of all 4 base colours in different percentages) with a 50% Noise in the colour stettings. However, all 5 files had been laid out totally the same – in fact the element containing that grey fill had been the "same" in all of them (just handed on via files created by "Save as…“ from the original first file.

The first thought had been that maybe one of the files had been damaged/corrupted in one way or another thus producing a different shade when printed, but when they printed all files again, it was a different one of those 5 files than before which was printing a different grey than the others.

So far my printer is clueless as to what may be reponsible for these inconsistent results from basically identical files/elements – they said they never had encountered something like this before…

Is there a possibility that the process of random generation of "Noise" of a certain percentage on a fill is somehow preserved in the PDF used for printing? Making the RIP produce slightly different shades in different print runs? I actually would have thought that these "Noise" fills were rendered as rastered pixel content when exporting to PDF… or don't they?

[NotMyFault]

Can you provide and example file (afdesign plus PDF) to reproduce?
opening the PDF in Affinity will show what has been rasterized or not.

[Lorox]

6 minutes ago, NotMyFault said:

opening the PDF in Affinity will show what has been rasterized or not

Yeah, I actually could have thought of that…

And yes, the rectangles with that grey Noise fill have obviously been rendered as images to their respective PDFs. And: when I copy one of those images to one of the other files and place it just over the image there in "Difference" blend mode it is showing 100% coverage (with just a bit of 99% flickering) in every single CMYK channel, which seems to prove that the images are virtually identical in each PDF...

The original grey colour is 23C/19M/18Y/45K with 50% Noise assigned.

I'm attaching one of the signs as .apub & PDF. As I said before: the other .apub ones had all been derived/copied from such an original via "Save as…" with just the number changed each time.

Meanwhile I almost tend to think that maybe there actually has been some issue with the (foil) printer at the printers' shop as in different print runs colours changed with different ones of the PDFs. When the grey images in the PDFs are virtually the same in any one of those PDFs I cannot see why they should print differently were it not for some flaw in the printer itself (or its RIP).

Sign_0_180x180+3_grey-NOISE.pdf Sign_0_180x180+3_grey-NOISE.afpub

[lacerto]

I had a look on your Publisher file and noticed that one of the noise swatches was K-only based while the other were CMYK-based. One possible explanation for different tones in different print runs is that you have previously had the grays produced in K-only, and subsequently in four colors. Within Affinity apps, this can happen very easily, e.g., if switching the color mode within Colors panel while having the lock (inadvertently) off, or when importing initially grayscale-based colors, which in Affinity apps get interpreted as composite grays, and would be converted in CMYK productions to four-color grays.

UPDATE: Note too that when exporting, simply just picking a color profile that deviates from the current document color profile, would result in K-only based grays to be converted to four-color grays (Affinity apps do not have an option to "keep numbers" at export time (assign a different color profile), so if there is need for that, it must be done by assigning a different profile using File > Document Setup > Color (or Document > Assign ICC Profile when using Photo).  

[lacerto]

One other consideration is related to the Noise generation itself applied. Has this previously been done outside of Affinity apps? At least the Live filter noise, even when monochromatic and applied with K-only gray, will generate noise also on C, M and Y channels. Or was this perhaps previously produced in grayscale or Gray/8, and then output as DeviceGray, thus keeping the output essentially monochromatic, rather than CMYK?

[Lorox]

@lacerto Thank you very much for your insights! Actually, those four RAL color swatches in the "HH Grautöne RAL-like" palette had been created first according to some info about the (approximate) CMYK values of several RAL-colour tones. And yes: according to the charts which I had consulted, the last in line of these swatches ("RAL 9007 Graualuminium") should be just 60%K and nothing else.

However, the guy at the printer’s was a bit anxious that a larger flat area in the "design"printed as monochrome tint of just ONE printing colour (K here) might be more delicate and any subtle flaws (should there be any) e.g. with the printing head for K might possibly result in visible "scanlines" or the like whereas an area of grey colour mixed from all 4 CMYK colours might be not so easily damageable.

So I created – just visually – a full CMYK equivalent of that 60K tint (resulting in 23/19/18/45). However, I forgot to add it this version as an extra swatch to my "HH Grautöne RAL-like" Document Palette and just used that mix directly on the square behind the number...

To further make it a bit more "fuzzy" I later added 50% Noise (with that slider in the "Colour" Palette) to it (the square, that is).

[Aside: Those 2 "pre Noise" versions of the colour – K only and full CMYK – are visually undistinguishable on my system, in Publisher as well as in e.g. PACKZVIEW. Also in Acrobat Pro on my old iMac. Provided, though, all apps are using the same CMYK color profile. If you add that 50% Noise the two versions of the colour will  look a bit different – but I guess that should be expected given their different "mix"! In "Preview"(not using profiles?), however, there is a visible difference, even with no Noise added.]

Supposedly simple things can obviously get quite complicated…

Anyway:
as those different print runs which I mentioned have been done with the very same PDFs of the files (which, as we have seen, eventually even have that grey area as a rendered and identical image in them) it still seems strange enough that the colours in the prints should visibly vary on each iteration and PDFs which printed this way the first time print differently the second time. I cannot quite see, what influence anything that has or has not happened before outputting the PDFs in Publisher should have at this point in the process… It's the same PDF that's beeing printed and yet the colours differ! So meanwhile I really tend to think – as I said before – it's probably something with the printer itself (or its RIP).

 

[lacerto]

Thanks for the extra information. Yes, having the exact same PDFs then obviously needs a different explanation!

Just for curiosity, I wanted to compare applying 50% Noise (I did not even remember this toggle in the Colors panel!) on K60 and on C23M19Y18K45

As can be seen, applying it on CMYK conversion does introduce some "color" (as expected).

Here is how applying Live Noise filter (via Photo Persona) distributes the noise (Gaussian and Uniform versions of Monochrome noise applied on K60 and its CMYK conversion):

The Live Noise versions seem to produce more even results, and might be worth a try in the future... 

The mere production variations seem endless, one does not need print time unpredictability to make any addition to that!

UPDATE: I forgot to mention the main thing: any kind of noise applied on mere K channel generates C, M and Y values (in addition of creating noise on K channel). Therefore I am not sure how beneficial it is to convert the initial neutral K value to CMYK components. It would be possible to delete C, M and Y output to make pure monochrome noise, but as seen from the curves above, having noise distributed more evenly also on C, M and Y channels (low levels) might make a clear visual difference on paper. 

Sign_0_180x180+3_grey-NOISE_variations.afpub

 

 

 

 

[NotMyFault]

27 minutes ago, lacerto said:

Here is how applying Live Noise filter (via Photo Persona) distributes the noise (Gaussian and Uniform versions of Monochrome noise applied on K60 and its CMYK conversion):

Take care when using live noise filter. If used atop a vector shape and grouped with default blend mode Passthrough, the noise will impact the alpha channel which you never want in a print document.

I raised this bug ages ago and it will probably stay forever.

 

[Lorox]

@lacerto Thanks a lot for your interesting take on the Noise subject. I wouldnt't have thought of going to the Live Noise Filter in the Photo Persona under these circumstances – it's certainly worth a try in the future!

[Lorox]

2 minutes ago, NotMyFault said:

Take care when using live noise filter. If used atop a vector shape and grouped with default blend mode Passthrough, the noise will impact the alpha channel which you never want in a print document.

Another thing I wouldn't have thought of… Thanks for pointing this out!