Copy link to clipboard
Copied
I was tracing a PSD file - template for box - thus fairly sparse graphics, high resolution image 8kx8k. It probably didn't need to be that res, but that is what I was handed. AI trace basically turned the file to mud no matter what settings I used. Now I use Trace a lot, so I was really surprised at the terrible tracing functionality. As a test, I cropped out a single graphic, take it to AI and traced - and it did fine, so I'm thinking that the algorithm compares details to file size and simple doesn't try very hard on small details if they are too small in comparison to the overall file size. I resided the incoming file to 50% - still junk. Graphics had detail that wouldn't survive further shrinking. I ended up using Image2Vector, a clunky mac app, that still required going to a half size file - but it worked well. It seems to me that AI needs another slider allowing the user to select some degree of detail - in return of course for lots more CPU time, but at least a reasonable result can be obtained.
{Renamed by MOD}
Copy link to clipboard
Copied
Hi there,
Thanks for reaching out and sharing this observation. I will definitely pass on this feedback to our Product team and would request you to post it on our UserVoice page (https://illustrator.uservoice.com/) as well.
If possible please attach the PSD file as well. That will help us replicating this issue and fix it accordingly.
Once done, please share the UserVoice link here, so that others can also upvote and share their observations and files. This is the best way of communicating with the Engineering and Product Management teams regarding issues and suggestions so they can be implemented in future releases.
Regards,
Srishti
Copy link to clipboard
Copied
Copy link to clipboard
Copied
As far as I can see, nothing changes about this. What I've learned is that AI controls the quality of conversion based on the size of the canvas. The bigger the canvas, regardless of complexity or square pixels of graphics, the lower the quality of tracing. AI could easily (and probably already does) break up the image by islands of graphics. It could convert them separately at high resolution. OR - allow the user to select a portion of a raster image and convert it - then an other portion etc. I don't believe either would create a huge programming workload (and I am a programmer). I continue at times, have to use a third party application that is slow and painful to do certain conversions and I'm just as sure I'm not alone.