@Kruskebunken this looks fantastic! I enjoyed reading your journey making this model, and it really shows the progress and the time put into it. The wood looks interesting. From the spec you can really feel the polish layer on top of the wood itself, like from this angle:
Great job!
YairMorr
Hey guys! This is my progress so far on the "Time Bomb" using Blender with Boxcutter
Pep_mepla
At the end of the day, all it takes to understand the infringement aspect of it (which, with a little bit of perspective gained over the last few weeks, is actually clear as day) is to replace the anthropomorphic term "learning" by the much more straightforward term "using".
It could be rather easily demonstrated by simply building two models : one using a certain amount of images and metadata + the Afghan Girl, and one without. And then the software just need to be prompted with "Afghan Girl" and the results compared. The distinction between the model using the picture without any consent and the one which isn't would be obvious.
As a matter of fact I would assume that it is much easier to demonstrate than what the art thieves are implying. And I believe there is not even any need for any new law, as copyright law already covers it all. The tipping point would only be the interpretation of the court as to whether or not the theft constitutes fair use. And at the end of the day the thing that determines what fair use is ... is the consensus around artistic practice. Artists are ok with showing their work without any watermark or restriction to access for other humans to appreciate it for free and get inspired ; they are not ok with the images being used without consent in an image generation algorithm. That's all it needs for it to be put into law if needed eventually.
As for going in circles : well, arguments for or against (either in good faith or as a devil's advocate) only go in circles if nothing new is being added to the conversation. So, after a point has been made, there really isn't much point repeating it ad nauseum. I think everyone is interested in hearing new ideas and original arguments, just not the same old Photoshop analogy (which IMHO barely applies here to begin with ...).
pior
At the moment this really only affects people who sell a finished piece . People who make art as part of a process will be insulated against the problem until the AI gets smart enough to interpret very specific feedback - which will happen eventually.
If you make a living off your style (lois van baarle is the best example I can think of) then an AI that can generate images in your style is a legitimate threat to your livelihood.
I'm of the belief that there should be regulation that makes commercializing a robot trained on specific artist(s) portfolio without authorisation a breach of some sort of IP/copyright protection. it won't stop it happening but it will at least give you somewhere to aim the lawyers when it does (amounting to roughly the same level of protection your work has traditionally had)
impossible to regulate huh. this will likely move the hands of many providers and websites to enforce this.
It's a nightmare for recruiting since now you have to distinguish between AI art and human art and not all of it is labeled.
I do like the fact that it's disrupted the trending algorithm since I never liked that aspect of Artstation.
That said AI to me really counts as spam filling what ought to be a professional space, so unless hiring catches up with recruiting prompt artists, all the AI garbage ought to have its own place elsewhere.
NikhilR
( --//SUFFIX TO ADD TO NEW LAYERS suffix = "_LP" --//FUNCTION TO GET OR CREATE A LAYER FROM A NAME fn getOrCreateLayerFromName layerName = ( newLayer = layermanager.getLayerFromName layerName if newLayer == undefined do ( newLayer = layermanager.newLayerFromName layerName ) newLayer ) --//GET THE CURRENT SELECTION selObjs = getCurrentSelection() --//CLONE THE NODES maxOps.cloneNodes selObjs cloneType:#copy actualNodeList:selObjs newNodes:&newObjs --//ITERATE THE SELECTION TO ORGANIZE THE NEW NODES INTO NEW LAYERS for o = 1 to selObjs.count do ( --//CREATE THE LAYER IF NEEDED oldLayer = selObjs[o].layer newLayer = getOrCreateLayerFromName (oldLayer.name + suffix) --//ADD THE CLONED OBJECT TO THE NEW LAYER newLayer.addNode newObjs[o] --//CHECK THE PARENT OF THE OLD LAYER pLayer = oldLayer.getParent() if pLayer != undefined do ( newParent = getOrCreateLayerFromName (pLayer.name + suffix) newLayer.setParent newParent ) ) select newObjs )
monster
Currently working with this character. Guys i need an advice. Is this looking like organic and fleshy or it is more like a rubber duck?
why would they be useless? it doesnt even have to be a human readable watermark. machines are perfect at hiding information in some off pixels here and there. or even with visible watermarks, it doesnt have to be crossing the entire image.
if its not murder, how about a seatbelt? man have people been upset when their freedom was taken to drive a car without.
or piors example in france, where you have to tell that a picture was manipulated and is in fact not a real photo.
that is very very close.
I actually think watermarks is a pretty good idea. It doesn't fringe on censorship or denial of services. of course, people can simply not use them or remove them, the same way people do now. but saying it will help nothing is going too far I think.
Gun laws, copyrights, banning of drugs, murder being illegal, etc definitely work. Of course not to 100%, but to a reasonable point that doing nothing would be far worse.
lotet