Key Links --------------- Humble Link: humblebundleinc.sjv.io/CorelDrawBundle Alt Link: bit.ly/3Abp3LH GFS Link: gamefromscratch.com/amazing-new-corel-draw-humble-bundle/ The above links give GFS a commission if used (and thanks so much if you do!) If you have trouble using the Humble Link try the alternate link. If that still doesn't work, paste it into a new browser tab. There is an issue with Humble Links and some adblockers. Timeline --------------- 0:00 Vector FX, Vision FX and HDR FX Overview 1:40 Hands-On With HDR FX 3:25 Hands-On With Vision FX 9:24 Hands-On with Vector FX 14:53 Conclusion and Outro
Yes, you can use Vincent van Gogh's style because his paintings are in the public domain and not copyrighted. This is because the artist has been deceased for more than 70 years. Steamboat Willie, the original Mickey Mouse, is also in the public domain now because the trademark ran out.
The COPYRIGHT for Steamboat willie has expired not the trademark (not sure they trademarked it of course just pointing out the trademark copyright) . Trademarks NEVER expire which is why Mickie Mouse eventually going intot he public domain won;t mean all that much as Mickey Mouse is still a Trademark for Disney. I believe trademarks can be abandonned and if a trademark becomes a common term it can be lost. For example Kleenex becoming the common term for tissue paper.
Those plugins are shovelware. But considering CorelDraw Standard alone costs about 200 EUR, getting it for 30 is a bargain, but then there are better programs out there (Illustrator, Affinity Designer).
The Corel Standard suite provided in the bundle is also crap. No CMYK and no pdf export at a resolution above 150dpi that makes it useless for printing.
Have to agree... it's all pretty awful. It's like having generative image AI from 3 years ago on your own PC. Hell, I tell Vector FX to give me a picture of a woman with short hair, and her hair is halfway down her back. I ask for a pony tail, and her hair is loose and free. I went into it with no illusions. I just figure that I can use it for easy placeholder images when I'm making something. Also because last I checked, Krita got confused cos my mouse settings are left handed and it didn't understand that the pen was a pen. I got Painter in an earlier bundle and I figured the cost of this one was reasonable for the upgrade anyway.
Quote from the FAQ section "What models do Vision FX and Vector FX use? Vision FX 2.0 uses Stable Diffusion XL (SDXL) Vector FX 1.0 uses SDXL Turbo and SDXL Vision FX 1.0 uses Stable Diffusion 1.5" Here you are. So it's stable diffusion.
@@gamefromscratch Yes true. But don't 1.5 specially referees to the checkpoint/weights? With algorithm/architecture being named just stable diffusion (v1)? The terminology is a bit unclear.
@@gamefromscratch I suspect that they're trying to whitewash it. If they're using Stable Diffusion, the base model is absolutely a derivative work of everyone and everything. They may well have created their own dataset to *customise* the model, but as far as I understand it, they're simply biasing a model to make things more similar to things in their dataset. This means that the output will not be visibly similar to the works it's derived from, but it's still derived from them.
Corel used to offer a substantial vector graph clipart package with each generation of their software, I imagine it would be more than enough to train their AI model.
If you have a strong enough GPU you can use Krita with AI Image Diffusion with ComfyUI. It's free and very powerful but just takes a little effort and requires a decent GPU (Nvidia 12GB or more or Apple Silica). You can give a prompt to aid with your drawing skills or generate layers as a stencil. Seems better then Vision FX.
Kinda wish you wouldn't promote AI tools unless it circumvents some actually bad aspects of game design. For instance (and I know this is a film example but it's the only one hat comes to mind), I wouldn't mind and AI that auto keys stuff out for me. But I don't want one that does the generates for me. "I want an AI that does my laundry so I can draw, not an AI that draws so I can do my laundry" -someone on twitter I saw idk
This ^ AI tools can be fun, like -woah, a computer can do that?- but at some point it's just young the creative stuff. I'd say coding counts as -creative- too.
That Humble Bundle was disappointing and if I had my time over I would not go with it. It was also negative PR for the suppliers, it feels like it under-delivered on what was implied in the offer. I wont look at anything from the suppliers involved there in the future. Still, Humble Bundles generally deliver, so not blaming them.
Curious if you have any experience with other vector generators. I’m interested in just tinkering and experimenting and would be interested in something that’s not complete garbage
I never tried to make money off AI pictures because I couldnt fix the mangled limbs, but it seems other people with lower standards are making bank off it. Feel real stupid now.
LOL. That "HDR" plugin is so full of false marketing. It's just a bunch of common color correction presets that people do in more professional software. There is no HDR going on and you're not turning a SDR image in sRGB gamut at max 100 nits to a HDR image in P3 or Rec.2020 gamut with highlights reaching up to 1000, 2000, or even 4000 nits 🤣🤣🤣
People do tend to get drawn into this sort of thinking easily... 1. I want to do good things. 2. I get money from this thing. 3. This thing enables me to do good things. 4. This thing is a good thing. But the problem we have today is that really no-one is paying the reviewers, so they have to put referral links, which ultimately means the people making the software are paying the reviewers, so they're not on the buyers' side....
This is what I’ve been waiting for to decide. Saw the past comment on the first video about Corel Standard version being locked down and limited. Distinct AI products would be the only thing I’d consider using. Let’s go
I get really sick of people constantly going off about "ethically sourced" artwork for these generative tools, because pretty much *no one* does the same when it comes to code-generating tools. The artists can go straight to hell as far as I'm concerned; most of them are nothing but hypocrites. Even legally speaking, there's nothing wrong with duplicating someone's style as long as you're not copying their direct works or claiming they made something that they didn't. The algorithms that get distributed don't contain the images used in their generation, so there's no copyright claim to make against them either.
@@lashghost Copying someone's style is long settled and is not protected. That's as close as "derivative works" gets in terms of art. You'd have to directly copy someone's specific piece of artwork in order to have a copyright claim against you, whether in their style or not. The only other copyright claim would be in regards to distribution of the artwork used to train the algorithm, and that's not how they work. You can't unpack or otherwise reverse engineer said artwork, so that angle isn't a valid legal complaint either.
@@kyliefire5008 You contribute nothing to the conversation, and probably even less more generally. Learn something before speaking next time; until then, be quiet while the adults are talking.
ai frequently barfs out unmodified chunks of people's copyrighted work, and plenty of software people aren't happy about their code being stolen either. you're believing their press releases uncritically.
@@zsheets7483 The algorithm is going to use reference images of specific artists, otherwise how can it reduplicate a specific style? Thereby that is a way to track where it made derivative works, despite not being able to reverse engineer the work like you would in hardware. But then again maybe it is possible to reverse engineer digital artwork. 3D models are all planes with mathematical coordinates on a X,Y, and Z axis. Along with this textures are maps again with mathematical calculations built in. Essentially 3D models are like code, which means they could be reverse engineered. In the case of 2D, it's normally done in layers in a program like photoshop and can be deconstructed. Again all of it is digital and can be reverse engineered on some level. The legislation just has not caught up to AI yet. You are incorrect in saying that copying someone's style is long settled and not protected as if you can prove in court that the person used your art to create derivative works in court, meaning you can look into the process and isolate where that happened, you could win a case. Meaning that as soon as someone types or puts in an image to the AI that says " I want a style that is like Nintendo Zelda", you have already entered into the territory of creating a derivative work of that specific copyrighted work. A person can draw from imagination but an AI doesn't really have that to fall back on as everything is a calculation of some type. Maybe in the future it'll be possible to have a program like that seen in the academic world where it can calculate what percentage of the art is copied/plagiarized and that could be taken into court, in the case of AI.
Most of these tools are just things you can already do for free with tools like Stable Diffusion only with a lot less control. To create good AI art you absolutely need tools like fine tuning, ControlNet and inpainting. People need to get over this "ethical" and "unethical" nonsense. When huge corporations such as Disney start training models on all their owned art and data, technically they will be completely ethical. People not using new tech because they think they have some morale highground isn't going to do them any favours when in the future they get replaced by someone doing their job ten times quicker because they are using AI.
@@Sprogster Yes unfortunately that's what a lot of AI stuff is like at the moment. They just take existing tech and repackage it with marketing buzzwords like being "ethical". I guarantee they haven't trained their own base models for this because it's still really expensive. Anyone can fine tune models but they won't technically be "ethical" unless the base model they fine tuning is "ethical".
They just made a use interface for stuff that already exists. Buuuut ComfyUI and the like also already exist, so why even care? That's like the people who are super excited to pay 20$ for some ai code editor which can already be replaced with other FOSS solutions for literally free. Like - "Woah, a product! It must be good, since it has a price!"
Key Links
---------------
Humble Link: humblebundleinc.sjv.io/CorelDrawBundle
Alt Link: bit.ly/3Abp3LH
GFS Link: gamefromscratch.com/amazing-new-corel-draw-humble-bundle/
The above links give GFS a commission if used (and thanks so much if you do!) If you have trouble using the Humble Link try the alternate link. If that still doesn't work, paste it into a new browser tab. There is an issue with Humble Links and some adblockers.
Timeline
---------------
0:00 Vector FX, Vision FX and HDR FX Overview
1:40 Hands-On With HDR FX
3:25 Hands-On With Vision FX
9:24 Hands-On with Vector FX
14:53 Conclusion and Outro
Yes, you can use Vincent van Gogh's style because his paintings are in the public domain and not copyrighted. This is because the artist has been deceased for more than 70 years. Steamboat Willie, the original Mickey Mouse, is also in the public domain now because the trademark ran out.
The COPYRIGHT for Steamboat willie has expired not the trademark (not sure they trademarked it of course just pointing out the trademark copyright) . Trademarks NEVER expire which is why Mickie Mouse eventually going intot he public domain won;t mean all that much as Mickey Mouse is still a Trademark for Disney.
I believe trademarks can be abandonned and if a trademark becomes a common term it can be lost. For example Kleenex becoming the common term for tissue paper.
you're doing gods work actually testing the tools you're covering. thank you!
Terrible bundle. Had to ask for a refund (which humble quickly provided). The AI stuff was a complete joke.
Those plugins are shovelware. But considering CorelDraw Standard alone costs about 200 EUR, getting it for 30 is a bargain, but then there are better programs out there (Illustrator, Affinity Designer).
The Corel Standard suite provided in the bundle is also crap. No CMYK and no pdf export at a resolution above 150dpi that makes it useless for printing.
Have to agree... it's all pretty awful. It's like having generative image AI from 3 years ago on your own PC. Hell, I tell Vector FX to give me a picture of a woman with short hair, and her hair is halfway down her back. I ask for a pony tail, and her hair is loose and free.
I went into it with no illusions. I just figure that I can use it for easy placeholder images when I'm making something. Also because last I checked, Krita got confused cos my mouse settings are left handed and it didn't understand that the pen was a pen. I got Painter in an earlier bundle and I figured the cost of this one was reasonable for the upgrade anyway.
How did you get a refund if you tried it already? You revealed the keys right? I didn't think they allowed that.
Thank you so much, you saved my time and money! Big love and happy new year!
Quote from the FAQ section
"What models do Vision FX and Vector FX use?
Vision FX 2.0 uses Stable Diffusion XL (SDXL)
Vector FX 1.0 uses SDXL Turbo and SDXL
Vision FX 1.0 uses Stable Diffusion 1.5"
Here you are.
So it's stable diffusion.
I might play around with this vector stuff, thanks for the models!
That's the algorithm though, not the training data.
@@gamefromscratch
Yes true. But don't 1.5 specially referees to the checkpoint/weights? With algorithm/architecture being named just stable diffusion (v1)?
The terminology is a bit unclear.
I honestly don't know, as you say its very unclear.
@@gamefromscratch I suspect that they're trying to whitewash it. If they're using Stable Diffusion, the base model is absolutely a derivative work of everyone and everything. They may well have created their own dataset to *customise* the model, but as far as I understand it, they're simply biasing a model to make things more similar to things in their dataset. This means that the output will not be visibly similar to the works it's derived from, but it's still derived from them.
Thanks for doing the more detailed review! 👏👏👏
Corel used to offer a substantial vector graph clipart package with each generation of their software, I imagine it would be more than enough to train their AI model.
On the other video, I mentioned that in the FAQs, they mentioned they are using the Stable Diffusion 1.5 model for VisionFX, at least.
So... is it considered ethical or not ethical? 🤔
How did you install these two plugins in Affinity Photo? They seem to have their own installers and go ... where?
If you have a strong enough GPU you can use Krita with AI Image Diffusion with ComfyUI. It's free and very powerful but just takes a little effort and requires a decent GPU (Nvidia 12GB or more or Apple Silica). You can give a prompt to aid with your drawing skills or generate layers as a stencil. Seems better then Vision FX.
Tried the vector fx, that worked ok but the vision fx was so slow to generate an image it wasn’t usable.
u need a cuda gpu for that ut works fine on my pc
Ai. Pass from me dawg.
Nah dude I'd rather learn how to make art quicker than use all this AI bullcrap...
Totally legit. I fully understand and appreciate that perspective
Everyone will be using AI, you'll be at a disadvantage.
@@scottcastle9119 I don't give a shit
@@scottcastle9119 Or maybe he'll be at an advantage, since everyone will be generating slop and his art will stand out...
@@thawne4235 are speaking in terms of what AI is right now? Because AI is a tool that grows every year faster than any software.
Thank you for your review.
"Buy once, create forever". The tagline loves by every designer.❤
That "Responsible AI" paragraph was a joke. Without telling us on what material they trained the AI, there's no way to verify those claims.
With Vector Fx can you import and trace images to vectors?
No, it's generative only
It's really bad. Essentially just a standard generator with a shitty post process vector conversion. Totally useless.
Kinda wish you wouldn't promote AI tools unless it circumvents some actually bad aspects of game design. For instance (and I know this is a film example but it's the only one hat comes to mind), I wouldn't mind and AI that auto keys stuff out for me. But I don't want one that does the generates for me.
"I want an AI that does my laundry so I can draw, not an AI that draws so I can do my laundry" -someone on twitter I saw idk
This ^
AI tools can be fun, like -woah, a computer can do that?- but at some point it's just young the creative stuff.
I'd say coding counts as -creative- too.
That Humble Bundle was disappointing and if I had my time over I would not go with it. It was also negative PR for the suppliers, it feels like it under-delivered on what was implied in the offer. I wont look at anything from the suppliers involved there in the future. Still, Humble Bundles generally deliver, so not blaming them.
bs this was a great hb bundle for 32 euro
Is that Chocolate in the Hamburger?
Vector FX is an absolute junk. Whatever you prompt it gives you garbage results compared to other AI vector generators
Curious if you have any experience with other vector generators. I’m interested in just tinkering and experimenting and would be interested in something that’s not complete garbage
@NightmareAndNinja I highly recommend ReCraft
I never tried to make money off AI pictures because I couldnt fix the mangled limbs, but it seems other people with lower standards are making bank off it. Feel real stupid now.
I think Flux does a good job of it.
Yet another software suite to avoid.
LOL. That "HDR" plugin is so full of false marketing. It's just a bunch of common color correction presets that people do in more professional software. There is no HDR going on and you're not turning a SDR image in sRGB gamut at max 100 nits to a HDR image in P3 or Rec.2020 gamut with highlights reaching up to 1000, 2000, or even 4000 nits 🤣🤣🤣
Maybe their AI tools have been trained on their own assets? I think that's something that Adobe does as well.
i thought adobe admitted to using their UGC in their AI
your humble link dont work
Paste it into a new tab, they changed it. There is no slider anymore, but don't worry it still helps me (and thanks)
JUST WOW 😂
Does he know he's an ad ?
It really seems like he doesn't. He's talking about this the same way like he talks about godot, which makes me uncomfortable
People do tend to get drawn into this sort of thinking easily...
1. I want to do good things.
2. I get money from this thing.
3. This thing enables me to do good things.
4. This thing is a good thing.
But the problem we have today is that really no-one is paying the reviewers, so they have to put referral links, which ultimately means the people making the software are paying the reviewers, so they're not on the buyers' side....
There is no AI going on anywhere, it does not exist yet. It is just ML.
This is what I’ve been waiting for to decide. Saw the past comment on the first video about Corel Standard version being locked down and limited. Distinct AI products would be the only thing I’d consider using. Let’s go
I get really sick of people constantly going off about "ethically sourced" artwork for these generative tools, because pretty much *no one* does the same when it comes to code-generating tools. The artists can go straight to hell as far as I'm concerned; most of them are nothing but hypocrites. Even legally speaking, there's nothing wrong with duplicating someone's style as long as you're not copying their direct works or claiming they made something that they didn't. The algorithms that get distributed don't contain the images used in their generation, so there's no copyright claim to make against them either.
Copyright affects more than just the original work, but also derivative works, which seems to be the major sticking point regarding AI.
@@lashghost Copying someone's style is long settled and is not protected. That's as close as "derivative works" gets in terms of art. You'd have to directly copy someone's specific piece of artwork in order to have a copyright claim against you, whether in their style or not. The only other copyright claim would be in regards to distribution of the artwork used to train the algorithm, and that's not how they work. You can't unpack or otherwise reverse engineer said artwork, so that angle isn't a valid legal complaint either.
@@kyliefire5008 You contribute nothing to the conversation, and probably even less more generally. Learn something before speaking next time; until then, be quiet while the adults are talking.
ai frequently barfs out unmodified chunks of people's copyrighted work, and plenty of software people aren't happy about their code being stolen either. you're believing their press releases uncritically.
@@zsheets7483 The algorithm is going to use reference images of specific artists, otherwise how can it reduplicate a specific style? Thereby that is a way to track where it made derivative works, despite not being able to reverse engineer the work like you would in hardware. But then again maybe it is possible to reverse engineer digital artwork. 3D models are all planes with mathematical coordinates on a X,Y, and Z axis. Along with this textures are maps again with mathematical calculations built in. Essentially 3D models are like code, which means they could be reverse engineered. In the case of 2D, it's normally done in layers in a program like photoshop and can be deconstructed. Again all of it is digital and can be reverse engineered on some level.
The legislation just has not caught up to AI yet. You are incorrect in saying that copying someone's style is long settled and not protected as if you can prove in court that the person used your art to create derivative works in court, meaning you can look into the process and isolate where that happened, you could win a case. Meaning that as soon as someone types or puts in an image to the AI that says " I want a style that is like Nintendo Zelda", you have already entered into the territory of creating a derivative work of that specific copyrighted work. A person can draw from imagination but an AI doesn't really have that to fall back on as everything is a calculation of some type. Maybe in the future it'll be possible to have a program like that seen in the academic world where it can calculate what percentage of the art is copied/plagiarized and that could be taken into court, in the case of AI.
Ive been using AI to create my own game dev tools.
First comment 🎉🎉🎉
Most of these tools are just things you can already do for free with tools like Stable Diffusion only with a lot less control. To create good AI art you absolutely need tools like fine tuning, ControlNet and inpainting.
People need to get over this "ethical" and "unethical" nonsense. When huge corporations such as Disney start training models on all their owned art and data, technically they will be completely ethical. People not using new tech because they think they have some morale highground isn't going to do them any favours when in the future they get replaced by someone doing their job ten times quicker because they are using AI.
Yep absolutely. The AI is just re-skinned free stuff that doesn't even work that well.
"People need to get over this "ethical" and "unethical" nonsense."
No.
@@Sprogster Yes unfortunately that's what a lot of AI stuff is like at the moment. They just take existing tech and repackage it with marketing buzzwords like being "ethical".
I guarantee they haven't trained their own base models for this because it's still really expensive. Anyone can fine tune models but they won't technically be "ethical" unless the base model they fine tuning is "ethical".
@@sinisterdesign Good argument, you've convinced me!
They just made a use interface for stuff that already exists. Buuuut ComfyUI and the like also already exist, so why even care?
That's like the people who are super excited to pay 20$ for some ai code editor which can already be replaced with other FOSS solutions for literally free.
Like - "Woah, a product! It must be good, since it has a price!"
Extremely bad software that is too expensive even at a discounted price. Anything above 0 is too expensive!