Preflight profile Convert to PDF/X-3 (SWOP) detects transparencies where there are none

[Robert24451006mfgi]

I'm trying to convert a PDF v1.6 to the X-3 standard (so downgrading it to v1.3, which doesn't support transparencies) using the SWOP profile in Preflight named "Convert to PDF/X-3 (SWOP)" which checks for transparencies and should flatten them automatically with the "Flatten transparencies" fix as you can see here:

After creating the new file the Results view in Preflight says that the only thing that doesn't meet the X-3 standard is transparencies, and then lists 1448 elements as transparent (see attached the Results generated as a PDF with images in the report compressed so I could attach it here) although I've successfully run the whole document through the Flattener Preview before attempting to convert to X-3, flattening the few images that actually had transparent backgrounds and later checking every page and seeing, that Flattener Preview didn't see transparencies anywhere on them (Basically I checked on every page, that Highlight could not be set to "Transparent Objects" nor any other option in Flattener Preview. I assume this means that there's nothing to highlight - no transparency on the page, because before flattening the few pages with transparency I was able to select "Transparent Objects" highlight on these pages).

I believe that the flattener worked and that these are false positives, because there's nothing transparent about any of these objects in the document. But sadly I can't generate an X-3 compliant document because of these false positives. 

I'd appreciate your help.

[JR Boulay]

I cannot reproduce this issue, can you share (part of) this document?

[Robert24451006mfgi]

@JR Boulay have you tried reproducing the issue with the report file that I attached? It contains all the same content that didn't pass the transparency check.

[JR Boulay]

You're lucky that I'm curious, I removed the 1448 comments and the first page.

Your report is about font transparency issues:

The report I got is not about transparency it's about missing glyphs, so there is obviously a font issue in this document:

Can't you export directly in PDF/X-3 from Scribus?

.

[Robert24451006mfgi]

@JR Boulay thank you for your effort. I didn't realize the comments were a problem. Here, I attached a compressed file that results in exact same report on X-3 conversion attempts.

I can't downgrade to X-3 with Scribus because Scribus can't flatten transparencies.

Strange, why didn't the font errors show up in my Preflight Results then? Seems like they're caused by the report generation. itself, because when I optimized the report file, I took care to uncheck everything except image compression.

[Test Screen Name]

The message (which is rather specific and complicated) does NOT say "fonts are not embedded". This is what the normal preflight checks for. The message says "the font IS embedded but does not have every character you use". This would seem to be an error with the PDF generator (scribus?)

[JR Boulay]

You should export to PDF 1.3 from Scribus, so the PDF will not contain transparencies.
Have you tried it?

In any case you should especially take a close look at all the texts that use the Oxygen font.

[Robert24451006mfgi]

@JR Boulay I've tried it. That's the reason I installed Acrobat DC in the first place - Because Scribus can't flatten transparencies. Scribus's developer recommends to flatten using GhostScript CLI or Acrobat DC.

[Robert24451006mfgi]

Here's a reference: https://github.com/aoloe/scribus-manual-pdf-and-printing/blob/master/content/transparencies.md

[JR Boulay]

Can you please share the "original" PDF, no processed by the Preflight fixups?

[Robert24451006mfgi]

@JR Boulay the file that I shared in my other comment hasn't gone through any Preflight fixups, but I did use Convert Colors and Flattener Preview on it. So here's the file as it comes from Scribus, except with compressed images. Hope that helps.

[JR Boulay]

1. Output Preview : Object inspector : Click on any text object : since it's a 100% black text overprint should be on (true).

2. The Flattenner Preview cannot show any transparency in this document, the "Highlight" menu is grayed out:

3. Clicking "the "Apply" button using these settings screws up all the texts:

4. Trying to convert this messed up document to PDF/X-3: no transparency issue but it returns a "Missing glyph" error and so it fails:

5.

==> I guess the transparency problem is a mirage and the real issue is the Oxygen font.

You should try to use another font, or another version.

[Robert24451006mfgi]

@JR Boulay thank you for spotting the problem with overprint=false. I will try running the "Set black text to overprint" Preflight fix after color conversion and before transparency flattening as soon as I get home. 

[Robert24451006mfgi]

@JR Boulay reporting back - Setting black text to overprint just made it worse. Now instead of having gibberish on pages that have images with transparency in them I get gibberish on all pages, as flattening now works on overprinting the text as well (adding the colors together instead of knocking the shapes of 100K black colored letters out of the background image).

Next I tried generating the PDF in Scribus without embedding the fonts in hopes that maybe I could do the embedding in Acrobat. All text in the resulting PDF became italic, so I think this approach is off the table.

Next I tried subsetting all fonts in Scribus. This prevents text turning into gibberish, but still registers a lot of false positive transparency in the X-3 generation report.

Tomorrow I will try substituting all Oxygen Regular with a common font like Arial, keeping it as an embedded font and see if that helps solve the false positive transparency error. 

[Robert24451006mfgi]

Here are the results of substituting Oxygen with Arial:

Transparency errors with Oxygen font:

Transparency errors with Oxygen font substituted for Arial in Scribus and then the same process repeated:

There are 52 less text-related transparency errors. Not sure why. The main difference is that Arial takes up less space than Oxygen, creating shorter text columns. Maybe Acrobat is splitting the text into a smaller number of text fields, so all text fields still throw an error, but there are now less text fields, so less errors. 

[Robert24451006mfgi]

Like so:

[Robert24451006mfgi]

Just to clarify on why in the accepted answer setting to overprint worked but in this comment it didn't is, because in the accepted answer I set all fonts to subset instead of to embed, which got rid of the problem of text turning into gibberish. 

[JR Boulay]

Can you send me (private message) the SLA file and the fonts?

[Robert24451006mfgi]

I did.

By the way, do you mind explaining why I should overprint the black text over the parchment background? Was it just a possible technical solution or is this something I should be doing in general?

[JR Boulay]

I installed the fonts, opened the SLA file with Scribus 1.5.8, exported to PDF 1.3, and converted to PDF/X-3 with Acrobat Pro without any issue.

See attachment.

[Robert24451006mfgi]

Hi @JR Boulay 
The thing is, I can't export to PDF 1.3 from Scribus, because 1.3 doesn't support transparencies. And as I've mentioned before, Scribus can't flatten transparencies, it instead just loses all layers except one in the process of generating a 1.3 compatible file. Here's an example:
Spot art with transparencies will have a fuzzy outline like this one:

When exported from Scribus as a PDF 1.3, it will look like this:

 Without this problem I wouldn't need to use Acrobat DC at all.

You'd see the same result if you tried exporting without deleting all the images first.

[JR Boulay]

"is this something I should be doing in general?"

Yes.

Otherwise you can have trapping issues.

[Robert24451006mfgi]

Thanks, I'll keep that in mind 😃

[Robert24451006mfgi · Accepted solution]

@JR Boulay I have good news 😃 I managed to get rid of practically all of the transparency errors in the X-3 conversion! Now only 6 matches instead of 895.

This is thanks to your recommendation to use overprint. I ran the fix called "Set black text to overprint" (under Preflight - Acrobat Pro DC 2015 Profiles - wrench button - Color, spaces, spot colors, inks) as the first step in Acrobat DC right before color conversion. (I also tried it after color conversion, but it created a greyscale copy of each illustration for some reason. I wanted to keep all images in the ICC profile specified in color conversion.)

After that I did color conversion, flattening, DPI reduction of images and conversion to X-3 just like I've done before, so the overprint fix is what made the difference.

I suspect that the remaining 6 matches are caused by the background image and a page number box being fully covered by a full-page illustration. I will try making a new master page for this specific page that has no background image and no page number box (even though the box is empty on that page) and repeat the rest of the steps.

[Robert24451006mfgi]

I was right about the remaining 6 errors - I changed the master page of page 2 and set the full page illustration there to overprint and there are no more transparency errors with this file during X-3 conversion: