2026-01-21 07:27 PM - edited 2026-01-21 07:53 PM
We are now on Archicad 29 and PDF creation is as painful as ever !
What is the point of having image textures (which are great for bringing the project to life on screen) when PDF output is so problematic? Currently trying to publish drawing sets and complete BIMx sets and every time the publication process hangs as soon as it reaches the first sheet of sections/elevations. We can, painstakingly, publish them one sheet at a time. However, every attempt to publish a set hangs (spinning beach ball). Perhaps it would eventually publish if left to run for a sufficiently long spell but that ties up the workstation.
I'm not sufficiently technically knowledgeable to know why Archicad PDFs are so large, cumbersome, & problematic, but its driving us insane...and its hardly a new problem. (layers & project info deselected, images & curves limited to 150dpi. Lossless images compression)
...A3 sheet with 4 sections of a small project hangs publication. Published individually it gives a 26MB file that both Preview and Nitro PDF Pro struggle to open.
Are we missing something? Doesn't everyone come across this problem? And, why on earth hasn't it been resolved yet?
Any advice that might help, please?
Edit:
Graphisoft: Please add the option to publish "flattened" PDF files to a user defined resolution. Would this not resolve a large part of the problem, and let us benefit properly from the "surfaces" in Archicad?
2026-01-30 07:55 PM
IIRC an option is to lower DPI during import.
Mac Studio M4 Max 64GB ram, OS X 10.XX latest
another Moderator
2026-01-30 10:21 PM
Do you mean, compress all images when exporting from Archicad? (you seem to have typed 'during import' so I'm not sure.
If that's what you meant, then it is not a solution. It isn't the resolution itself which causes the problem (although it does affect matters to a degree). Rather, it's the complexity of the document that Archicad produces: it includes *everything* underneath/behind what we see, i.e. any hidden elements/textures etc, they are all there, 'behind' the final layer which we see on the screen (and which the printer will actually put to paper).
What the Ghostscript command will do is strip all that and then optimize a bit the remaining data. After the initial installation of Ghostscript itself (a simple paste of the command into Terminal and hitting Enter), the process is not complicated at all. I successfully processed many PDFs which presented particular difficulties, and for initial file sizes up to 10-15MB the process was very quick. I did also process an A1 drawing with multiple textured elevations and sections and, predictably, that one took about an hour on my 2018 MacBook Pro.
2026-01-31 12:15 PM
I see multiple parallel problems that converge when publishing final documentation.
1) Drawing complexity: I see no other reliable workaround than tracing over live model views on a worksheet being intentionally sparse with the amount of elements (in some cases turning transparency and model view to have no fills helps especially when the model is small). For historic buildings there is no other viable way, the required geometric complexity cripples even editing and modelling on decent machines - forget about your average Joe opening those PDFs on their potato.:)
The methods of blending 3D projections with hand drawn 2D can be argued, I did end up over the years with many experiments to keep my projects sharp and buttery smooth… stacking multiple views and cropping drawing boundaries on layouts, annotating on different layer combinations or design options to filter the model, keeping the structure/MEP as a reference drawing locked under the model view, completely separating publishing from modelling, basically ending up doing the work twice - but the core of the issue is Archicad crippling to a halt after a point due to its geometry being arcane.
Compare the same building modelled in Archicad or Rhino what is the result in a PDF. Same with Revit or Parasolid modellers. Hell, even Vectorworks! Until this level is not face lifted, idk how much can be salvaged by trying to outsmart it.
2) Image fills (or high-end imaging workflows in general)
I gave up on Archicad, and do these kinds of plans in Affinity. Does this mean more work? Yeah, but I don’t have an urge to throw my M1 Max MBP out of the window. If you have clean facades anyway (see 1)), this is quite liberating to have fluid control over textures, colours, shading, you name it.:)
Same goes for rendering, just get the model out to C4D or Blender, get the images rendered, grade them, then compare those files with the PDFs that have those images as external drawings. Ugh.
I keep my drawing set intentionally dry looking, at least it is up to my thinking and the flow of designing is not killed by the software. Let’s skip colour management which Archicad has 0 tools for.
For presentations, just save yourself the headache (…typography entered the discussion) and use a DTP software - Affinity is free since last October, spin it around. For submittals keep the drawings as minimal as possible, you don’t want to smile at the official while your emailed pdf is opening for 5 minutes and you don’t have a printed set. 🫠
3) Compliance: PDF/A remains a problem that I cannot circumvent with Affinity. In my country all submitted plans need to be PDF/A and Affinity does not support this format. I had a very expensive competition submission due to re-saving the package with Acrobat Pro when I learned this.
Reading heavy PDFs
For people on macOS, an extremely fast PDF reader I found is Skim: https://skim-app.sourceforge.io
I use Preview, PDF Expert, Acrobat on a regular basis depending on the project, but Skim chews through anything I throw at it.
Is this a solution? Absolutely not, but this is what I regularly end up with for more complicated geometries and bloated models.
+1
Vector PDFs are not needed to measure from them, I use Morpholio Trace to sketch over WIP plans - even screenshots can work as you just specify two points and set their distance and the software calibrates the canvas. I think even revision diffs can be rendered with image recognition, so I see little arguments against completely flat or reasonably high DPI image saves (with proper control!)
Where you would really need complexity in the PDF if you were to store the source IDs and exported properties (like Bonsai generates SVG drawings - look into that, pretty cool), bit AFAIK Archicad PDFs are not “smart” in this sense either.