Scribus' Rendering Question

Previous topic - Next topic

NFG

Hallo!  I'm a relatively recent arrival on the Scribus scene, but so far I'm loving the crap out of it.  It is a lot like my old DTP experience with Calamus and other software back when 32MHz was obscenely fast.  Anyway.

There's a lot to learn on any new app so I've been spending a lot of time on IRC with the very helpful crew there, but recently one of them said something which is a little bit alarming, and I need clarification about Scribus' rendering pipeline.  Specifically, he said that a 'spline' object, by which I must assume he meant vector objects like fonts and SVG drawings, were rendered to the DPI of the bitmap beneath them.

QuoteIf you have a semi-transparent spline base image atop of a raster, the printing conversion will have to rasterize the shape (type for example) to the raster quality.

Which effectively means things like type will rasterize to say, 300DPI versus the 1200DPI for text.

Now, for many reasons, this is utterly insane thinking, because if it is true it means I will have to prepare every bitmap image to the DPI of the output device (printer) just in case a vector object is ever placed on top of it.  I've been unable to find any documentation about the rendering process for Scribus so I'd really like to get some advice on this. 

I'm 50 pages into a book on video game graphics and I'd guess that the number of low-resolution sprites I've added numbers in the hundreds.  Better to know now I've wasted my time than another 150 pages from now!  =D

Nermander

I don't think this applies only to Scribus, but printing in general. The text says it applies only to semi-transparent objects (which implies transparency flattening, and that is something Scribus does not yet inlude).

So, a black vector on top of a bitmap image will not be rasterized.

NFG

Thanks.  Yeah, it does seem to only apply to semi-transparent objects, but boy, I am utterly baffled by the logic behind it.  At what point does it make sense to render a vector object based on the completely arbitrary DPI of the raster object behind it?  What happens when the vector object is above two or more rasters with different DPI?

At this point it doesn't seem to matter, I don't use transparent or semi-transparent vector objects for this project.  But I might one day, I'm fairly proficient with inkscape.  This sort of loony restriction just does my head in.


Nermander

#3
It would be easy to test. Draw a 50% transparent hairline vector line above a 25 DPI bitmap image and look at the result?


So I just tried. Took a picture of a flower (from Openclipart.org), downsized to 50x50 pixels. Imported to an image frame, set the DPI to 25.

Then I drew a line on top of it, diagonally, hairline, solid, green, 50% transparency.

Exported to PDF and printed, the green hairline is still shown as hairline.

When printing directly from Scribus the green line didn't even print, not sure why, I have never printed directly from Scribus before.

NFG

I was told it's a final-stage rendering issue, so it won't show up in PDFs.  You won't know until you wast part of a tree.  =/

utnik

Quote from: NFG on May 21, 2014, 08:30:18 AMThis sort of loony restriction just does my head in.
take a look at other layout programs like indesignand see how transparency is flattened there!
often the standard is half the resolution of the image below and you have to tell the flattener, if you like full image resolution...

Quote...I'd guess that the number of low-resolution sprites I've added numbers in the hundreds.
...and you intend to put a small semitransparent text on top of this lowres stuff?
even with highres outlines, the big pixels inside the characters will make them almost unreadable – i would never put a small semitransparent text over an image!
...and for things like shadows, the resolution of the image is the highest possible without adding 'guessed' image content.

utnik

NFG

Haha, no, the fact that this applies only to text that's less than 100% opaque only really became known later.  I'm not completely deranged, I won't be putting faded text over images any time soon.  I put up a worst-case PDF if you'd like to see what I'm up to:

http://nfgworld.com/files/AssaultSuits-sample.pdf

Lots of PNGs with alpha transparency, no transparent text.  But still, it's a weird limitation.  One thing I'm learning about layout now is that I really, really did have some amazing tools in 1987.   =)

Nermander

As I wrote, I "exported to PDF and printed".

utnik

#8
hi nfg

your uploaded page doesn't contain anything with native transparency – so nothing will be flattened with the limitations of the ppi of the underlying image...

btw. i don't think, there was a software back in 1987 wich could handle transparency in a better way...

utnik

NFG

QuoteAs I wrote, I "exported to PDF and printed".

Sorry, I mis-read that the first time, thanks for clarifying.

Quoteyour uploaded page doesn't contain anything with native transparency

No, and I hope to keep it that way.  The way this was first explained to me was -all- transparent objects, so all these PNGs would be real trouble for me.  I'm glad to hear this isn't the case.  =)

Quotebtw. i don't think, there was a software back in 1987 wich could handle transparency in a better way...

I was just talking generalities, not this particular problem...  I'm surprised how little progress has been made in the two and a half decades since then, but that's not to say I don't love Scribus to bits.  I'm very keen on the new versions, I'll be first in line when someone gets around to a Windows release.  =)