Jump to content

Paul Mc

Members
  • Posts

    258
  • Joined

  • Last visited

Everything posted by Paul Mc

  1. Hi @drstreit Here's clip from the project page (https://waxy.org/2022/08/exploring-12-million-of-the-images-used-to-train-stable-diffusions-image-generator/) for that dataset: Also, I didn't see where you can go to have your images excluded do you have a link for that. The copyright issue is discussed briefly on the comments on that page but there are no answers.
  2. I'd figured that out but thanks for letting me know I'm sure it will be useful to anyone else who goes down the same pathway. I now realise that I was "blinded" by the fact that I was in the Move Tool [V] mode all the time so this possibility never appears in the UI and is therefore not discoverable.
  3. Hi @lacerto that's perfect, thank you. You learn something every day!
  4. Hi everyone, I've just hit a snag with a project I'm working on where I appear to be unable to snap guides to nodes on the curves of holes. My (original) workaround was to create a rectangle that snapped to the hole geometry and then snap the guide to the rectangle but now I've a bigger job with non-rectangular holes and this is getting a little tedious. The blue guide will snap to the left edge of the pink rectangle but not to the edge of the rectangular hole (in a yellow circle). The edge shows in purple indicating that it is a candidate for snapping. Turning off bounding box snapping results in the ruler not snapping to anything. Indicating that previously the ruler was only snapping to the bounding box and not the geometry. The pink rectangle will snap to the hole nodes/vertices and curves/edges though! I'm guessing that guides use a subset of snapping and align differently compared to when you are moving other types of objects around. The online help is a little fuzzy on this and uses the terms objects, curves and vertices in the description but doesn't elaborate on any differences with guides as far as I could see. This isn't a showstopper as I can use other objects/shapes for alignment. However, using guides would have reduced the risk of unexpected shapes ending up in the final print-ready artwork. So, am I doing something wrong or is this just the way it works?
  5. I would be interested in seeing some AI features too. I would much prefer an API to allow me build my own plugins though as this feels to me like a better use of the development resource. In addition to Affinity products I also use Blender extensively and I've recently been using DALL-E to to perform more creative in (and out) painting to textures and material images. Source and reference images occasionally arrive with too severe a crop or with noisy objects spoiling the composition. This feature of DALL-E is very helpful in extending sets/scenery and removing objects. Although the AP inpainting feature is very useful for large areas it can often get stuck in a rut with high contrast changes and clone copies appearing which means you have to use other approaches like cloning or painting. I also have a frequent need for seamless textures which often require lots of fiddle and faff in AP. Creating these in DALL-E (after some pre-processing in AP) nearly always gives me the best results. As a separate issue, I think that the industry is changing. AI is not going to go away. I'm already seeing marketing departments getting excited at (the potential of) not having to hire me to do some of the work they need. I, personally, feel that excitement is a little premature and that my skillset is still going to be required to complete the pipeline. Currently, the diffusion method that is popular isn't doing much to help with vector designs. My opinion is that these tools will help with inspiration and create a new kind of idea sketching that will be upstream of what designers and artists will receive as part of a brief and may even be part of their toolkits. It has already helped me with the blank canvas block syndrome for getting started after receiving intangible wording for an idea from the client. AI is part of the designer/artist future and we need to embrace it and understand it. That's not to say that it is good - a lot of what I've seen is rubbish, but it is getting better at some things. It is early days and things are developing quickly, plus there are still lots of problems. There are huge legal and ethical issues surrounding the sourcing and curating of the source training data. The web might be public access but that doesn't make all the images visible on it CC0!
  6. Yes! It has not happened to me recently but I think it's only a matter of time. Separate to this is that a recent project had me managing a complex tree of design alternatives and tweaks across several product briefs which were unique but had some elements of consistency. It was really hard work when each change request came through. Reviewing it and working backwards it was easy to see what I should have done 😊. I have a good backup system running here but even that felt too infrequent to save me should there have been a failure - or a mistake.
  7. Thanks @Andreas Scherer I knew about this feature but hadn't considered using it in this context. I've used it where there is a mainline of development with minor tweaks which may need to be removed It seemed to work OK but the final file didn't allow easy traversal and review of all the "final nodes" of the design without exhaustively cycling through all the decision points. For the moment I'm trying out the @walt.farrell method above and suffer the disk hit as I think that will be the most robust in the face of client request changes.
  8. Thanks @LondonSquirrel I hadn't considered the use of something like bsdiff - until now. I've used a patch builder under Windows many years ago and can see that there are several offerings on github with equivalents so I might investigate that option. I'm Windows based so ATM I'm not sure how your discussion of the cp command could be applied. I've just acquired a new QNAP NAS so disk space isn't a problem and QTS is a close relative of Linux so maybe I could push that functionality across to that box. It's early days on that one and I need to do some more homework before I'd feel comfortable having my paid work rely on it. Thanks for the divergence - it's all interesting stuff.
  9. Thanks @walt.farrell I might just give this a try. This is what I was thinking of doing when I posted the original message but was curious about how others might do this.
  10. Thanks @Pšenda, there isn't a problem with using SVN - my point was coming from a software developer perspective where branches and merges are a clear benefit which appear impossible with current offerings. And, yes, to use for snapshots is really the only use I can see, which, of course, will make the database grow.
  11. I've recently had the experience of a complex pathway of repeated refinements dealing with several stakeholders resulting in a set of final designs using AP &B AD. I've also been checking the forums and there appear to be a growing number of corrupt file saves being reported. Thankfully I've not had that happen. I'm just wondering if there is scope for a setting to keep a number of generations of backups of a design file, getting bumped on each save? I was doing this manually but every so often I'd accidentally save over the current version and this had me wondering if whatever is causing the corrupt saves hit me I'd be in a real pickle. Version control would be even better but all the systems out there wouldn't be able to detect changes other than checksum/size/datestamp which would make it a more tedious job (IMO) to maintain. Is anyone out there successfully using version control with Affinity products? Is there any other strategy to consider other than using a well defined folder/file naming scheme?
  12. @Harnaak I think I may have misunderstood the problem here. The font made no difference to what I was seeing. Here's how I turned off the ligatures: The Typography panel can be found under the Character tab/studio. Toggling the fi button is all that was required.
  13. The Calibri ligatures export fine for me. Here's a capture from Adobe Acrobat Reader: I don't have the GurbaniAkharHeavy font installed so I can't check if this has a side-effect.
  14. Maybe if you have a solid white coloured background on which you place the images that you might not see the dashes? Perhaps that's how MS Pub works? While in the application it is possible that the dashes are not visible on screen because they are transparent but when they are printed the printer driver converts transparent areas to black? Just a guess.
  15. If you try a rectangle over the whole area so that the edges of the rectangle align closely with the dashes and then use a clear/transparent fill but a white stroke colour and then on the stroke tab give the stroke a non-zero thickness. Maybe use a bright colour to start with and then switch to white once you can see where the edges are and it is in the correct place.
  16. Further to my answer, if the dark spots are due to dark pixels then only cropping will apply (unless you want to paint them out using the brush or maybe a white rectangle stroke with a clear fill). In my case the dark dashes were due to transparent pixels and putting the white rectangle behind them made them match the white paper so were then invisible so the eye.
  17. In one case I cropped it just as you say you did earlier, in all the other cases I placed a white rectangle behind it and then rasterised the result - this allowed me to enlarge the canvas a little which helped with my requirement.
  18. I've had a similar problem with scanned images (via Photoshop). It happened because the images needed rotating for the lines of text to be horizontal as the scans were a little skewed. The rotation seemed to have added extra columns and rows of pixels around the border in PS. Because some of them were anti aliased they appeared as semi-transparent over the black background I had. The dashed effect was due to the skew correction. None of this was caused by Affinity apps but they allowed me to correct the problem.
  19. The original PDF may have set the document colour settings to Grey so any imported images are converted to monochrome. If you go to File | Document Setup... and click on the Colour tab make sure that the Colour Format setting is RGB (or CMYK if this is for print) before inserting/placing the colour picture.
  20. I have no idea, sorry. What card do you have? I would be surprised if the card itself is a problem. I think you would have noticed bigger problems before now if that was an issue. I considered reporting the issue when I experienced it but it proved impossible to reproduce while screen recording which made me wonder if OBS (screen recording software) might be interfering with it. Since then it has happened rarely so I've been able to live with it. Sorry, this isn't very helpful is it.
  21. Hi @NAR, I have this need also. In my world these are called Title Blocks. For now I use a linked PDF file created separately. Of course this could be edited in AD but in my case I wrote an application for the job. Using a linked file means that the format can be setup once and the contents not accidentally edited while you are working on the design. Also, when I update the contents AD detects that (if the file is open, or the next time I open it) so the details are always up to date.
  22. I've a suspicion that it could be display driver related. If it is on then maybe try turning off hardware acceleration in the Performance tab of Preferences and see if that helps.
  23. I've also seen this effect. It's a little strange and as you say not easily repeatable. My guess is that when you update the design the underlying canvas gets "invalidated" which queues up a redraw to render the update. Sometimes the redraw isn't completely redrawing the canvas and some parts retain their old content. Eventually those parts are redrawn or the whole canvas is redrawn and the problem goes away. Are you seeing old bits of the design lying around after you move objects?
  24. If I understand correctly it is a case of deleting the curve that represents the hole in each profile. If you double-click on a shape then it will select all the curves in it and also switch you into using the node tool. Then select one node on the hole curve; press Ctrl+A and then Delete.
×
×
  • Create New...

Important Information

Please note there is currently a delay in replying to some post. See pinned thread in the Questions forum. These are the Terms of Use you will be asked to agree to if you join the forum. | Privacy Policy | Guidelines | We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.