Gaps between vector shapes?

[true_blue1878]

I used AI to create a vector image, but when i import it into affinity I see hairline white gaps between each shape, which are also apparent when then saving it as a png or jpg. The gaps don't seem to be there in AI, so is it an export issue with AI or something with affinity thats doing it? 

Tried exporting from AI as both svg and pdf. 

Cheers! 

[firstdefence]

It's very likely Affinity has broken your image up into small images. Have you used a gradient mesh or something Affinity doesn't have and consequently cannot translate accurately?

[true_blue1878]

9 hours ago, firstdefence said:

It's very likely Affinity has broken your image up into small images. Have you used a gradient mesh or something Affinity doesn't have and consequently cannot translate accurately?

Something very simple. Just a low colour crown. 

As you can see with the purple velvet area, AI did a good job at vectorising it, but then when it gets exported and opened in affinity, it has a hairline gap between most of the shapes!

It could be something not right with my export settings in AI, I'll keep playing with the settings, I have a feeling it could be related to the "decimal" value with the export options, but for some reason, affinity is taking AGES to open the SVG files, so it's a difficult one to solve with trial and error. 

There is also an option in auto-trace to "overlap shapes", but I'd rather not.

Cheers!

[lacerto]

If it is all vectors and you intend to export to raster formats, you should be able to avoid those antialiasing stripes by selecting the violet shapes and turning off antialiasing by using the layer Blend Options (or by trying a different coverage map curve):

This does not have effect on vector based exports, but in print context those white gaps normally will not show. If turning off / adjusting antialiasing results in too rough edges, one option is to supersample the exported raster output (first upsampling using nearest neighbor and then downsampling using e.g. Bicubic resampling).

[true_blue1878]

7 minutes ago, lacerto said:

If it is all vectors and you intend to export to raster formats, you should be able to avoid those antialiasing stripes by selecting the violet shapes and turning off antialiasing by using the layer Blend Options (or by trying a different coverage map curve):

This does not have effect on vector based exports, but in print context those white gaps normally will not show. If turning off / adjusting antialiasing results in too rough edges, one option is to supersample the exported raster output (first upsampling using nearest neighbor and then downsampling using e.g. Bicubic resampling).

Ahhh I think I understand, so basically it is just either AI exporting with antialiasing or affinity antialiasing, but the printer obviously won't do that so there won't be any gaps when it's printed? 

I'll check out the blend options cheers!  

[lacerto]

8 minutes ago, true_blue1878 said:

but the printer obviously won't do that so there won't be any gaps when it's printed?

Normally the resolution is so much bigger when printing so the gap is kind of an exaggeration of the issue when viewing on screen (and typically shows at certain zoom rates and and vanishes in other) -- but if you want to be sure, there might be point in placing underlying objects with the same color under those stripes to make sure they vanish (this might require much manual work but may be the only way to avoid those gaps when producing vector designs that are viewed on screen, and viewed at close range).

[true_blue1878]

3 minutes ago, lacerto said:

Normally the resolution is so much bigger when printing so the gap is kind of an exaggeration of the issue when viewing on screen (and typically shows at certain zoom rates and and vanishes in other) -- but if you want to be sure, there might be point in placing underlying objects with the same color under those stripes to make sure they vanish (this might require much manual work but may be the only way to avoid those gaps when producing vector designs that are viewed on screen, and viewed at close range).

Yes was thinking that too, just drawing underneath to match the colour of the shapes. I will do that and then leave it as an optional layer and see if the printer wants to use it. Thanks mate! 

Also one more question.. is the AI auto-trace good enough for sending to printers? I find that when trying to trace complicated rasters (I.E. A face), it can create over 1000-3000 shapes just on the low-end setting of 20% colour. I have tried combining shapes but then when I click "join" it says they are not joinable layers  Am I best just leaving complex stuff like faces as raster? 

Cheers!

[lacerto]

8 hours ago, true_blue1878 said:

is the AI auto-trace good enough for sending to printers?

i am not sure if I understood what you mean, and additionally my experience of AI autotracing is mostly limited to working with the old CS6 version. It is not clear what actually is causing those artifacts when you open an autotraced document in an Affinity app -- most typically these kinds of issues happen when there are adjacent objects with the same color, and they are pretty common in all vector graphics apps. AI can avoid them in certain situations, e.g. when using outside aligned strokes, and it can also do some "autotrapping" (repositioning and adding undercolor) to avoid the issue. These kinds of artifacts could also be a result of having used gradient meshes (as mentioned by @firstdefence) which fails because Affinity apps do not support the feature, but if the shapes have been created by using tracing, this does not seem to explain the cause.

[true_blue1878]

4 hours ago, lacerto said:

i am not sure if I understood what you mean, and additionally my experience of AI autotracing is mostly limited to working with the old CS6 version. It is not clear what actually is causing those artifacts when you open an autotraced document in an Affinity app -- most typically these kinds of issues happen when there are adjacent objects with the same color, and they are pretty common in all vector graphics apps. AI can avoid them in certain situations, e.g. when using outside aligned strokes, and it can also do some "autotrapping" (repositioning and adding undercolor) to avoid the issue. These kinds of artifacts could also be a result of having used gradient meshes (as mentioned by @firstdefence) which fails because Affinity apps do not support the feature, but if you the shapes have been created by using tracing, this does not seem to explain the cause.

Sorry, what I mean is... 

Is simply letting AI auto-trace an image into vector curves going to be acceptable to a printer? I was worried because it seems to split solid colours up into lots of odd little sections, even though the exact same colour is directly next to it. Like, even with a black outline I drew around my design, the auto-trace split that line up into about 100 shapes!

The image totalled at about 3000 shapes and 40,000 anchor points lol. Will a printer accept that, or is it just too messy for them to use?

Just messed with this example image to show what I mean.. I am having similar issue with my design where it's creating too many shapes and finding too many shades of black when the rasterised colour is just pure black! Though most of them are clustered next to each other and the same colour. 

 

 

 

Is this kind of vectorisation going to be too complicated/heavy for the printer to print? Cheers!

[v_kyr]

That's a common problem (...those white boundary antialising lines, like small outlined puzzle peaces) with Affinity software when opening vector files which are generated by some tracers/vectorizers. Those can be limited and often successfully removed the way lacerto already showed/told above.  - But a vector trace of an image in contrast to the photo image, will always be visual recognized as being an artificial trace. Meaning here, why do you want to let then a traced image be printed here, instead of printing the sharper (probably without having any distortions) bitmap image?

 

 

[lacerto]

3 hours ago, true_blue1878 said:

Is simply letting AI auto-trace an image into vector curves going to be acceptable to a printer?

This is a difficult question to answer. Complex objects with lots of anchor points might fail to print but I do not think that there are absolute limits that can be stated that are common to all printers and jobs, so proof printing would be required.

It seems that the your sample images (at least the latter one) have been traced with "High Fidelity Photo" or with an equivalent preset, which kind of reproduces photo-like appearance but with vector shapes, which might be useful in some situations, but on the other hand, the first image has been simplified by reducing the number of colors (producing clear tonal zones).

Perhaps the antialiased stripes were actually results of manually "combining" adjacent shapes by making their colors identical, but without actually combining the shapes to single curves, since this is exactly the situation where such gaps are likely to occur. If so, the recommended method is to lower the number of colors already when tracing so that each tone comprises a separate shape.

[true_blue1878]

20 hours ago, v_kyr said:

That's a common problem (...those white boundary antialising lines, like small outlined puzzle peaces) with Affinity software when opening vector files which are generated by some tracers/vectorizers. Those can be limited and often successfully removed the way lacerto already showed/told above.  - But a vector trace of an image in contrast to the photo image, will always be visual recognized as being an artificial trace. Meaning here, why do you want to let then a traced image be printed here, instead of printing the sharper (probably without having any distortions) bitmap image?

 

 

Well, the printers customer service lady speaks very simple English, and they only use AI. So it's been difficult sending my design to them as AI imports Affinity designs in the worst way possible. It scatters and splits layers up randomly, it separates text into individual letter shapes and removes effects and without having access to AI I had to try and read between the lines of what she was asking for. I am not quite sure if she wanted just a vector outline for the white ink to be laid, or whether she requires the image to be completely vector. I am going to try sending them the flat image first and see if they can print that over the white spot colour underlay.  I am hopeful this will be acceptable to them. But if not, then I will have to try and convert the images to vector, but I have zero experience or knowledge of how to do that. So it will be a huge task considering the images are real faces! I have had a play around with auto-trace and I feel like to get anything near resembling photo realism, I'd need over a thousand paths and tens of thousands of anchors per face. It just feels so messy. 

16 hours ago, lacerto said:

This is a difficult question to answer. Complex objects with lots of anchor points might fail to print but I do not think that there are absolute limits that can be stated that are common to all printers and jobs, so proof printing would be required.

It seems that the your sample images (at least the latter one) have been traced with "High Fidelity Photo" or with an equivalent preset, which kind of reproduces photo-like appearance but with vector shapes, which might be useful in some situations, but on the other hand, the first image has been simplified by reducing the number of colors (producing clear tonal zones).

Perhaps the antialiased stripes were actually results of manually "combining" adjacent shapes by making their colors identical, but without actually combining the shapes to single curves, since this is exactly the situation where such gaps are likely to occur. If so, the recommended method is to lower the number of colors already when tracing so that each tone comprises a separate shape.

Yes I think it is about finding the sweet spot for each image. But there are more than one setting, so it is definitely a learning "curve". I'd rather not have to do it with these sorts of images. They're just too complex. But in future I'll definitely be making my more simpler designs with full vector in mind! 

Just waiting to hear back from the printer about my latest sample submission, which is just the raster image on top of a white underlayer set to "Overprint". I bit the bullet and paid for AI and made sure it was fully ok in there. In doing that, I learnt how absolutely awful AI is at importing other apps work. I really wish printers used Affinity, it is superior in so many ways. People are designing fancy logos and things in Affinity, sending it to their printer, and they open it up in AI which dissects/removes all effects and screws with the layers so much. Then they're told it can't be printed and people then have to go and recreate it from scatch in AI but without most of fancy effects Affinity can do. It's really frustrating.

 

Will update soon. Thanks all!

[firstdefence]

One a the questions I'd ask about tracing raster is, what are the advantages of vector in these instances?

[lacerto]

1 hour ago, true_blue1878 said:

I learnt how absolutely awful AI is at importing other apps work.

I am not sure that I understand what you mean. As for vector formats, AI can open and save to pretty much any old AI versions (just consider that Affinity 2.x apps cannot save to 1.x Affinity versions), open and  place PDF (open more complete feature set than Affinity apps do and fully support placed PDFs), and open and place standard (e.g. Corel exported) EPS (including spot colors, overprint attributes and embedded fonts), and SVG. It additionally opens older versions of CorelDRAW .cdr, and supports to some extent AutoCAD formats. Later versions might be even more compatible -- what is mentioned here is based on the CS6 version.

I am not sure if native formats of other apps have been made public, so it is not surprising that AI cannot open e.g. Affinity or Xara formats. The same is true for native AI -- some apps (like Xara) support certain AI specific features to some extent, but most vector graphics apps only support the PDF stream parts of AI, and non-AI specific parts of Illustrator exported EPS files that have Illustrator editing capabilities included. There simply just does not exist a universal vector file format that is designed for transferring features and document structure as fully editable. SVG can be pretty compatible but then it is limited to RGB and basically aimed at digital production.

Another thing is that many of the fancy effects that can be created in Affinity apps, cannot necessarily be exported without rasterization in any of the production file formats supported by Affinity apps, so to produce in high quality, you are often required to rasterize in high resolution. Direct printing capabilities of Affinity apps are pretty limited, as well, so even if print shops did have Affinity apps installed, they could not necessarily print Affinity created designs successfully to devices from within Affinity apps.

Many kinds of productions are still relying on a workflow of opening PDF, AI and EPS files in Adobe Illustrator, and then printing from within there, and these file formats are sadly typically ones that are supposed to have been created from AI and contain all native AI specific editable features. AI is basically the only app that can include this information.