Can you create a video about how you actually install Stable Diffusion checkpoint? Really want to try this out and went to hugging face and so confused about what I need to do.
Not sure what biggest thing you mean. You can see a screenshot of the exact path where they have to go at 2:09, with little red arrows and everything. You just copy-paste the files you downloaded in the 'models' folder. That's it.
Really cool! It would be helpful to explain how to install controlnet, or point to a different tutorial on how to do it. I have controlnet, but the different Control Net Units are not showing. Those tabs are not available. I can only see "Single Image" or "Batch"
any idea on why I don't have the different unit ? I use controlNet 1.1.22 EDIT : you need to go to settings to increase ControlNets from 1 to 2 at least I guess
@@user-bn7pw9tc4g go to settings, and then stable diffusion, show all then, CTRL+F and write controleNet, there will be two "ruler" the first is set on 2 or 3, and the second on 1, set it on 2 and save.
All QR codes shown are scannable. A lot more to show but wanted to keep the video relatively short. Might make a Part 2 soon. Note: If you're having trouble scanning some of them, try moving back a lot or make the screen very small. I made them huge in the video so you can see them better, which can cause a scanning issue. Check more QR code art on our twitter. twitter.com/MakeTheBreakOG All pertinent information is in the video description as usual!
@@vintagegenious no, the point is - it depends on the hardware/software. Mine couldn't read it. The qr-codes are supposed to be reliable. Making them pretty but unreliable is a bad idea unless until you know how many of your potential customers will walk away frustrated and angry at you because you couldn't just make a readable qr-code. (or whomever you have instead of customers - fans, subscribers...)
@@gdegde4406 I scanned all of them multiple times. Including after posting the video. All work fine prnt.sc/gKENtB_9fW1U Try moving back like 1.5-2m before scanning, they are stretched out in the video and are 10x the size they would normally be.
@@Makethebreak yeah, and are you going to tell this to your customers after they leave frustrated? Look, I'm not asking you to fix my problems with your qr-codes. I'm saying it's a bad idea to use them because instead of reliable message one can read easily in any lighting and circumstance you get a pretty message that some (you don't know how many) people won't be able to read without jumping through hoops. That is if they even know through which hoops they need to jump. And you would not be there to tell them (otherwise you probably would not need the qr-code in the first place).
i do love the concept of these artistic qr codes, however even with your shown examples at different sizes my iphone 8 has problems picking most of them up, only about 20% working, and when i generate my own using your shown method i also have quite the problems, even when correcting via control weight, so maybe my old af iphone has a different qr code spotting code than newer iphones, who knows, next test will be as a print version instead of directly screen scanning, thank you for your instructions :)
Yes I made them huge on screen for you guys to see them better but when scanning they have to be about the size of the standard QR code, so you have to either move far back or make it small. As far as your results go, you need to fine tune the controlnet values, it took me a couple of hours of fine tuning until I got it just right on my specific model. After that 95% of my outputs were scannable.
Sure Aiden, you got the toolbox and know-how, you have to find the job. If you offer a product or service you can use these to bring people to your page or store. If you don't have that, then you can create very attractive ones and offer the service to businesses or marketing companies in your area and show them how their marketing will improve.
Hey ! Thanks for this, im trying since yesterday but I dont get the images blended just 3 Dif pictures, but not blended with the qr, im doing it with the google collab, its maybe for that?
Cool!. I just have a question, why do you use the preprocessors in both control nets? In this case they do nothing and could be left out. I have the same results with or without them
Use for example android QR scan function (in the top bar menu), as the camera might pick up a face and then not scan the QR code. So conclusion would be to not use faces inside the QR art. my2cents
Faces are fine as long as you're not too close, when you're far back the camera will spot the code easily and not see the face. If you're scanning too close up or the image is too big, the opposite will happen. The size of the QR code has to be about the size of the typical code when scanning for best results. So either move far back or make the image much smaller.
The brightness model is clutch. Was experimenting with tiles and canny nets, but had problems to get real images to pop out of the code, if they were supposed to stay scannable. Thanks for sharing this.
On the other hand I've seen people create the image of snowy village, with amazing 3D effect, that was a scannable QR code, so even in this video, it's not using its full potential.
For sure. Its just a tiny sample of what you can do. I've only used a couple variations to show the workflow. I've made awesome things besides the ones shown. You can try different models and prompts and get all kinds of cool results.
I followed everything to a T and used all the exact models as you but for some reason when I enable the tile model it wont show any QR code just an image, ive copied the exact settings you have and even tried different one. I cant get any different results.
Sure, I'm only using that one prompt in the video I can fetch it tomorrow from A1111 and paste it here in comments for you. Though there's nothing special about it, it's quite generic.
hello, can you please tell me what is the second network in ControlNet for? does the second network affect the rendering speed? other bloggers do not use the second network and it turns out beautiful too, is it mandatory? i just downloaded Stable Diffusion and do not understand. I use the translator, I hope it will be understandable
How can I locate the control net model folder? You mentioned that after downloading the two files, we need to put them into the control net model folder, but you didn't provide instructions on how to do so.
I just did it again, worked for me. prnt.sc/gKENtB_9fW1U They are very big on the video, try either moving back a lot before scanning of make the window small it should work.
Very cool! Have you tried using an image from a 3rd party i.e. like a logo not created within stable diffusion but rather imported into img2img? I tried but am failing miserably... images created within SD look fantastic though!
Hi Mark, I have not tried img2img I might do that sometime this week if I get a chance. In fact, I found 2 different ways to make these in txt2img, but explaining both ended up being a 15min video, so I cut one out, cause I noticed people on youtube barely watch 3min on average, they like short videos. I might end up doing a part 2 if I see people enjoying this video.
@@Makethebreak I noticed when you do it the QR codes that you inputted are generated a long with the special qr codes. Mine doesn't do that. So i'm thinking you have a setting set that we don't have by following this tutorial
I'm getting a controlnet error "Exception: Error while deserializing header: HeaderTooLarge" When I look it up most people fixed it by redownloading the controlnet extension cuz they got the wrong file. That didn't work for me. I have the yaml file and the py file in the "extensions\sd-webui-controlnet\models" folder and downloaded both again just to be safe. I reinstalled controlnet as well. I have an RTX 4080 but part of the error says (p, unet, model, lowvram). I'm pretty stumped. I really appreciate advice, but I don't expect any. We're all out here busy with life and picking our battles lol.
yes you have to play around with the parameters, some models tend to be more difficult than others too. Edit: Also, a tip, when you scan try moving back before scanning, like 1.5m back. The software on your phone is used to small sized squares, when its huge it struggles. It should represent a very small part of your phone screen. Basically is should be the size of a QR code. :)
I noticed under generate there is a Styles drop down , and you have Styles1 where do you get this? I dont have anything in my drop down available to select, and mine are not coming out as good wondering if this has anything to do with it
No no nothing to do with it. When you find a certain style you like you can save it and then load it to get a similar style. But that should be irrelevant to you. Nothing to do with generating the actual QR codes.
I have applied the same params as shown, but the images generated are without the QR, I have enabled the control net. Should i try with the checkpoint you are using?
@@Makethebreak Thanks for the reply, it was memory error actually, got a help from my friend to use his pc and follow the video, thanks! Also, any idea on how to approach img to img diffusion, basically I was exploring to combine qr with some brand logo. Thank again!
why doesnt it show the Control Net in my Interface? i also have different folders from yours so i dont really know where to put the 2 files, should they go to the Models folder?
Hi Andreas, at 2:09 I show the exact path where the files have to go. As for not seeing controlnet in your interface perhaps you didn't install it originally? You have to install controlnet under the Extensions tab and download controlnet models.
This is gonna sound super naive but does this require some sort of program? I’m trying to generate a good looking QR code for my wedding, have no experience in this field, and got lost at the “control net model folder” step.. Any and all help is greatly appreciated!
Hey I followed the setup on my macbook pro 15 with i7 processir and all the models liad fine but in text to image when I give it a prompt the output i get is just the image of my text prompt without the QR as base. I added QR as png in controlnet canvas trued number of times not sure why is it not using QR image as base to add the text prompt output over it. Ratger it produces it separately, what am I doing wrong?
@@skipper8005 I myself did not get a chance to try out if this will work but don't do txt to img. Try out img to img and load your QR in one controlnet 0 and your artistic image in controlnet 1 and QR code in img to img canvas.
I have this in the console: ERROR: The WRONG config may not match your model. The generated results can be bad. ERROR: You are using a ControlNet model [control_v1p_sd15_brightness] without correct YAML config file. ERROR: The performance of this model may be worse than your expectation. ERROR: If this model cannot get good results, the reason is that you do not have a YAML file for the model. Do you know where to get it by any chance?
Make sure you move back a lot cause they are too big in the video. Either that or make the window small, I just scanned all of them again to double check. All work. prnt.sc/gKENtB_9fW1U
@@Makethebreak I know! I went back to that spot several times. What I don't see is where to access the path. Granted I'm new at this but I couldn't figure out if this is a path within the program or on your computer. So sorry for the silly question but I spent an hour on this. It doesn't help that I'm new to stable diffusion. Thank you for your response.
Yep, tried all 15-20 appearing in this video after creating them, before making the video and even after making the video. Make sure you move back a lot cause they are too big in the video. Either that or make the window small. The phone software is used to a tiny code, you cannot scan a giant QR it wont recognize it. They scan fine: prnt.sc/gKENtB_9fW1U
I could not scan any of your qr-codes. Maybe it's software-dependent. The point is - so far it's a bad idea to actually use this. You have no idea what percent of your potential customers would be able to scan those what percent will walk away frustrated.
I have no issue scanning them. That said, You have to be at an acceptable distance for the phone to be able to scan. You cannot scan a huge image up close your camera software will not see the code. That is, on your phone cam the QR code has to be the size of a typical QR code, shouldn't take up more than 10% of the screen when you're scanning it. Otherwise your phone starts looking for other details within the image, like faces etc.
Thanks for the video! I'm having an issue with mine although I've followed the exact parameters specified in the video. I'm getting wacky colors and no sign of any image behind the QR code. I've adjusted the settings quite a bit and am not seeing an improvement. Any ideas on what could be going on here? I'll try to post an example of the issue. Thanks! Screenshot: drive.google.com/file/d/1H2pannxyAIGxODsevgDl4ZQWhGjaVkC2/view?usp=sharing
Your 'pixel perfect' parameter should definitely be off. You have that enabled. As far as the rest goes the parameters in my video are specific to revanimated2, that's the model I was testing with. It took me like 2 hours to get it balanced right. If you want to use another model you're just going to have to do a lot of trial and error till your parameters are perfect.
Wow, I never would have thought that they could be scannable. Some are beautiful and impactful. It already gave me a very good marketing idea.
Fast and clear explanation. Thank you :)
Can you make one for me, please
Can you create a video about how you actually install Stable Diffusion checkpoint? Really want to try this out and went to hugging face and so confused about what I need to do.
My man was explaining everything perfectly, then 2:06 hit and he was like, let me just skip the biggest thing you need to know. You got this.
Not sure what biggest thing you mean. You can see a screenshot of the exact path where they have to go at 2:09, with little red arrows and everything. You just copy-paste the files you downloaded in the 'models' folder. That's it.
@@Makethebreak We need to have stable diffusion webui downloaded first right?
@@goldenmidasfoot Of course :)
Really cool! It would be helpful to explain how to install controlnet, or point to a different tutorial on how to do it. I have controlnet, but the different Control Net Units are not showing. Those tabs are not available. I can only see "Single Image" or "Batch"
Woah! i cant belive they work holy.. super neat
Kindly make one for me
any idea on why I don't have the different unit ? I use controlNet 1.1.22
EDIT : you need to go to settings to increase ControlNets from 1 to 2 at least I guess
I'm looking around for the same answer!
@@user-bn7pw9tc4g go to settings, and then stable diffusion, show all then, CTRL+F and write controleNet, there will be two "ruler" the first is set on 2 or 3, and the second on 1, set it on 2 and save.
great video! Is there a way to convert an existing picture into a scannable QR code?
BLESS YOU!! I couldn't figure out how to do this after a few hours of trying.
Can you make one for me, please
Very nice method. If you could get that to work with external images in img2img that would be very cool, I'd like to see a part 2 :)
yes, would love to see that too!
Works great. Different models need a bit of playing with parameters as you mentioned, but it's an amazing idea! Thanks for sharing.
Welcome Brian, glad you enjoyed it :)
Can you make one for me, please?
@@vikrantvaishnav2867 In theory I don't mind at all, but I'm not even sure how to share contact info on youtube....
All QR codes shown are scannable. A lot more to show but wanted to keep the video relatively short. Might make a Part 2 soon.
Note: If you're having trouble scanning some of them, try moving back a lot or make the screen very small.
I made them huge in the video so you can see them better, which can cause a scanning issue.
Check more QR code art on our twitter. twitter.com/MakeTheBreakOG
All pertinent information is in the video description as usual!
>All QR codes shown are scannable.
No, they are not. I tried scanning 4 of them and it didn't work out.
@@gdegde4406 they are
@@vintagegenious no, the point is - it depends on the hardware/software. Mine couldn't read it.
The qr-codes are supposed to be reliable.
Making them pretty but unreliable is a bad idea unless until you know how many of your potential customers will walk away frustrated and angry at you because you couldn't just make a readable qr-code.
(or whomever you have instead of customers - fans, subscribers...)
@@gdegde4406 I scanned all of them multiple times. Including after posting the video. All work fine prnt.sc/gKENtB_9fW1U
Try moving back like 1.5-2m before scanning, they are stretched out in the video and are 10x the size they would normally be.
@@Makethebreak yeah, and are you going to tell this to your customers after they leave frustrated?
Look, I'm not asking you to fix my problems with your qr-codes. I'm saying it's a bad idea to use them because instead of reliable message one can read easily in any lighting and circumstance you get a pretty message that some (you don't know how many) people won't be able to read without jumping through hoops. That is if they even know through which hoops they need to jump.
And you would not be there to tell them (otherwise you probably would not need the qr-code in the first place).
i do love the concept of these artistic qr codes, however even with your shown examples at different sizes my iphone 8 has problems picking most of them up, only about 20% working, and when i generate my own using your shown method i also have quite the problems, even when correcting via control weight, so maybe my old af iphone has a different qr code spotting code than newer iphones, who knows, next test will be as a print version instead of directly screen scanning, thank you for your instructions :)
ok so with my ipad air it works a lot better with your qr codes, mine not so successful
Yes I made them huge on screen for you guys to see them better but when scanning they have to be about the size of the standard QR code, so you have to either move far back or make it small. As far as your results go, you need to fine tune the controlnet values, it took me a couple of hours of fine tuning until I got it just right on my specific model. After that 95% of my outputs were scannable.
Great video BUT how are you getting 2 ControlNet's??
you can set how many you want in the settings under controlnet (up to 10)
@@lukas5220 Thank you! I did find that after some digging around. Unfortunately I’m getting an error when I try to generate with ControlNet
I'm gonna need that negative prompt
We're early so any ideas on how to monetize this and get in on the tech. It's really cool
Sure Aiden, you got the toolbox and know-how, you have to find the job. If you offer a product or service you can use these to bring people to your page or store. If you don't have that, then you can create very attractive ones and offer the service to businesses or marketing companies in your area and show them how their marketing will improve.
Hey ! Thanks for this, im trying since yesterday but I dont get the images blended just 3 Dif pictures, but not blended with the qr, im doing it with the google collab, its maybe for that?
Really cool!
yes was waiting for this!
Is it possible to use your own image inlaid into the QR code? For example, a company's logo?
That 'should' be possible via img to img. I havent had to chance to play around with that yet, I'll try it this week.
@@Makethebreak i have tried it, doesnt work. let me know if it works for you.
@@Makethebreak I'm curious about this too
Amazing!!
Cool!. I just have a question, why do you use the preprocessors in both control nets? In this case they do nothing and could be left out. I have the same results with or without them
Use for example android QR scan function (in the top bar menu), as the camera might pick up a face and then not scan the QR code. So conclusion would be to not use faces inside the QR art. my2cents
Faces are fine as long as you're not too close, when you're far back the camera will spot the code easily and not see the face. If you're scanning too close up or the image is too big, the opposite will happen. The size of the QR code has to be about the size of the typical code when scanning for best results. So either move far back or make the image much smaller.
nice man
The brightness model is clutch. Was experimenting with tiles and canny nets, but had problems to get real images to pop out of the code, if they were supposed to stay scannable. Thanks for sharing this.
Why Google lens do not read this qr code
I tried this with some LoRas and the results were mixed :/ without them it worked like a charm!
Everything works, but my QR codes dont seem to scan in the images. What settings do you think tweaking would help with this?
On the other hand I've seen people create the image of snowy village, with amazing 3D effect, that was a scannable QR code, so even in this video, it's not using its full potential.
For sure. Its just a tiny sample of what you can do. I've only used a couple variations to show the workflow. I've made awesome things besides the ones shown. You can try different models and prompts and get all kinds of cool results.
I followed everything to a T and used all the exact models as you but for some reason when I enable the tile model it wont show any QR code just an image, ive copied the exact settings you have and even tried different one. I cant get any different results.
Strange. You got both controlnets enabled I take it?
What are your ideas on adding the prompt(s) in the description for us lazy people?
Sure, I'm only using that one prompt in the video I can fetch it tomorrow from A1111 and paste it here in comments for you.
Though there's nothing special about it, it's quite generic.
great since reddit is trash now this video helps
hello, can you please tell me what is the second network in ControlNet for? does the second network affect the rendering speed? other bloggers do not use the second network and it turns out beautiful too, is it mandatory? i just downloaded Stable Diffusion and do not understand. I use the translator, I hope it will be understandable
Any way to automate this process ? Using prompt not using UI
So cool.
There is any other way to create these types of QR codes.
Because stable diffusion is not working easily??
There is ill show it in the next video Monday or Tuesday. But the results are nowhere near as cool as stable diffusion.
@@Makethebreak thankyou,
i will be waiting 😊
I scanned one. I'm sure other models of phones will do it better... but mine blows.
Problem I have is that I cant selecte multiple control networks.. any idea?
How can I locate the control net model folder? You mentioned that after downloading the two files, we need to put them into the control net model folder, but you didn't provide instructions on how to do so.
I did. At 2:09 I posted a screenshot of the exact location of the controlnet models folder, with red arrows indicating the folders they are in.
respekt man, how do i solve the timeout google colab issue now ? then i will go all in with that stuff...❤
Been trying to find a workflow ever since the qrbtf guy made it. But I wasn't able to scan any of the first 3 QR codes in your video.
I just did it again, worked for me. prnt.sc/gKENtB_9fW1U
They are very big on the video, try either moving back a lot before scanning of make the window small it should work.
Very cool! Have you tried using an image from a 3rd party i.e. like a logo not created within stable diffusion but rather imported into img2img? I tried but am failing miserably... images created within SD look fantastic though!
Hi Mark, I have not tried img2img I might do that sometime this week if I get a chance. In fact, I found 2 different ways to make these in txt2img, but explaining both ended up being a 15min video, so I cut one out, cause I noticed people on youtube barely watch 3min on average, they like short videos. I might end up doing a part 2 if I see people enjoying this video.
Anyone else's controlnet having no effect on the output whatsoever?
make sure your models are downloaded and placed in the correct folder, also make sure the controlnet is enabled.
I'm having the same issue. I double checked everything
@@Makethebreak I noticed when you do it the QR codes that you inputted are generated a long with the special qr codes. Mine doesn't do that. So i'm thinking you have a setting set that we don't have by following this tutorial
I figured it out, you the the yml file downloaded for the tile thing as well
I'm getting a controlnet error "Exception: Error while deserializing header: HeaderTooLarge" When I look it up most people fixed it by redownloading the controlnet extension cuz they got the wrong file. That didn't work for me. I have the yaml file and the py file in the "extensions\sd-webui-controlnet\models" folder and downloaded both again just to be safe. I reinstalled controlnet as well. I have an RTX 4080 but part of the error says (p, unet, model, lowvram). I'm pretty stumped. I really appreciate advice, but I don't expect any. We're all out here busy with life and picking our battles lol.
what should I do when QR codes arent reading? anyone?
I can obtain very cool pictures but not scannable with this method... Will try to modify the parameters a bit, anyway great example of art
yes you have to play around with the parameters, some models tend to be more difficult than others too.
Edit: Also, a tip, when you scan try moving back before scanning, like 1.5m back. The software on your phone is used to small sized squares, when its huge it struggles.
It should represent a very small part of your phone screen. Basically is should be the size of a QR code. :)
I noticed under generate there is a Styles drop down , and you have Styles1 where do you get this? I dont have anything in my drop down available to select, and mine are not coming out as good wondering if this has anything to do with it
No no nothing to do with it. When you find a certain style you like you can save it and then load it to get a similar style. But that should be irrelevant to you. Nothing to do with generating the actual QR codes.
Are these Models safe? I been hearing of potential dangers with certain models.
I have applied the same params as shown, but the images generated are without the QR, I have enabled the control net. Should i try with the checkpoint you are using?
Both controlnets enabled with the same parameters as I have? and you downloaded both models etc?
@@Makethebreak Thanks for the reply, it was memory error actually, got a help from my friend to use his pc and follow the video, thanks!
Also, any idea on how to approach img to img diffusion, basically I was exploring to combine qr with some brand logo.
Thank again!
@@devpratap Did you find a fix for this on your own pc? It seems like if i upload a picture or not, it does not change anything.
why doesnt it show the Control Net in my Interface? i also have different folders from yours so i dont really know where to put the 2 files, should they go to the Models folder?
Hi Andreas, at 2:09 I show the exact path where the files have to go. As for not seeing controlnet in your interface perhaps you didn't install it originally? You have to install controlnet under the Extensions tab and download controlnet models.
This is gonna sound super naive but does this require some sort of program? I’m trying to generate a good looking QR code for my wedding, have no experience in this field, and got lost at the “control net model folder” step..
Any and all help is greatly appreciated!
HI Miyamoto, yes this tutorial assumes you have Stable Diffusion and controlnet installed on your PC. If you don't, you will need to do that first.
Hey I followed the setup on my macbook pro 15 with i7 processir and all the models liad fine but in text to image when I give it a prompt the output i get is just the image of my text prompt without the QR as base. I added QR as png in controlnet canvas trued number of times not sure why is it not using QR image as base to add the text prompt output over it. Ratger it produces it separately, what am I doing wrong?
Did you ever find a solution to this? It seems like if i upload a picture or not, it does not change anything
@@skipper8005 I myself did not get a chance to try out if this will work but don't do txt to img. Try out img to img and load your QR in one controlnet 0 and your artistic image in controlnet 1 and QR code in img to img canvas.
Dear blogger, why do I generate a special gray tone, the value I use is the same as you
Not sure Bai, maybe its the model you are using? or a special VAE?
I'd like to see this using my own pictures.
Is it possible to use this with dreambooth, or img2img?
I actually have not tried that, Ill give it a shot when I have a moment.
Can we do something similar with midjourney ?
Not to my knowledge. You do need controlnet to guide the AI.
I can't scan them 😥
I have this in the console:
ERROR: The WRONG config may not match your model. The generated results can be bad.
ERROR: You are using a ControlNet model [control_v1p_sd15_brightness] without correct YAML config file.
ERROR: The performance of this model may be worse than your expectation.
ERROR: If this model cannot get good results, the reason is that you do not have a YAML file for the model.
Do you know where to get it by any chance?
i'm also got the same error, does anyone know how to fix this?
where do i find control net folder
How come its generate 6-7 samples, while mine is generating only 1
You can pick the amount of samples under batch
what's the website link for the model selection?
They are in the video description section. Unless you mean SD in general those are on the civitai website.
Doesn't work for me, not scanable
Edit: second one works
Edit2: 3rd one doesnt, neither the 4th
tired on multiple monitors and different sizes
Make sure you move back a lot cause they are too big in the video. Either that or make the window small, I just scanned all of them again to double check. All work. prnt.sc/gKENtB_9fW1U
#QRDiffusion!
how do you access the controlnet model folder?
Hi Cynthia I showed a screenshot of the path to the folder at minute 2:08
@@Makethebreak I know! I went back to that spot several times. What I don't see is where to access the path. Granted I'm new at this but I couldn't figure out if this is a path within the program or on your computer. So sorry for the silly question but I spent an hour on this. It doesn't help that I'm new to stable diffusion. Thank you for your response.
@@growyourleads Even i am facing the same issue couldnt find it.
@@VaibhavOp me too i cant find it
"theyre scannable" did you even try before you made this video? they do not scan.
0:57 is the only one that actually scans
Yep, tried all 15-20 appearing in this video after creating them, before making the video and even after making the video. Make sure you move back a lot cause they are too big in the video. Either that or make the window small. The phone software is used to a tiny code, you cannot scan a giant QR it wont recognize it.
They scan fine: prnt.sc/gKENtB_9fW1U
I could not scan any of your qr-codes.
Maybe it's software-dependent.
The point is - so far it's a bad idea to actually use this. You have no idea what percent of your potential customers would be able to scan those what percent will walk away frustrated.
I have no issue scanning them. That said,
You have to be at an acceptable distance for the phone to be able to scan. You cannot scan a huge image up close your camera software will not see the code.
That is, on your phone cam the QR code has to be the size of a typical QR code, shouldn't take up more than 10% of the screen when you're scanning it. Otherwise your phone starts looking for other details within the image, like faces etc.
Yeah, cool
What's the purpose again? this still looks 90% ugly qr code and barely scannable. What's up with people's not making life easier.
i dont have inpaint_global_harmonious ;l and i cant find it
Thanks for the video! I'm having an issue with mine although I've followed the exact parameters specified in the video. I'm getting wacky colors and no sign of any image behind the QR code. I've adjusted the settings quite a bit and am not seeing an improvement. Any ideas on what could be going on here? I'll try to post an example of the issue. Thanks!
Screenshot: drive.google.com/file/d/1H2pannxyAIGxODsevgDl4ZQWhGjaVkC2/view?usp=sharing
Your 'pixel perfect' parameter should definitely be off. You have that enabled. As far as the rest goes the parameters in my video are specific to revanimated2, that's the model I was testing with. It took me like 2 hours to get it balanced right. If you want to use another model you're just going to have to do a lot of trial and error till your parameters are perfect.