ztmidgo / android-stable-diffusion-onnx Goto Github PK
View Code? Open in Web Editor NEW使用Android手机的CPU推理stable diffusion
使用Android手机的CPU推理stable diffusion
Hello developer, can you please add exif features in this app to automatically save metadata to the saved output images. Like the, prompt, negative prompt, sampling method, resolution, guidance scale, steps, seed, and the model name used to generate the image.
你好,想根据你提供的方法把整个app生成流程跑通。我对安卓apk部分完全是小白,能请问下载了github代码,有了模型文件后,要怎么生成apk吗?谢谢!
Hi there. I was successfully able to use fastsdcpu on termux to generate with android locally (rupeshs/fastsdcpu#123)
And just saw your app. First of all what is the most recent version of apk you have that is available ? The default apk you shared on the bottom just has one button on top and I presume that is move assets to memory and the other one is to generate.
And on one of the discussions you mention being able to load models externally , how would that be possible ? Where do we put them and how does the app see them ? How do we use other models ? (saw the other models on huggingface and was able to download a few of them)
你好,“下载 https://huggingface.co/TIEMING/Chilloutmix/tree/main 模型,复制模型到PathManager类中的getModelPath给出的路径”这部分,这个huggingface里有4个文件夹,每个都是模型的一部分,那么是要把哪个放在getModelPath给出的路径里呢?
First of all thank you so much for the new update.
Thank you for hearing out our feedback and implementing the save button and sampler option. Kudos to the developer.
Issue :
I think something is wrong with Euler Sampler. It's giving a totally different output for the same seed and prompts.
Example
Second output (same seed and prompt)
In comparison, DPMS sampler
First output
Second output (same seed and prompt)
Hoping for an update fix for this and with more samplers, imgToimg.
Suggestion1: Please add random seed to appear in seed edittext when onclick "Generate button" if "seed=null", so that we can know which seed was used and can recreate it again.
Suggestion2: To save some storage space, if possible, can you please make the app get model files directly from the phone storage while initialisation/generating instead of making a copy of the model files at app directory.
After the 2023.9.20 update I am facing crash issue while initializing on my Redmi Note 4 mido. The previous version was working without crash.
Here is the screenshot and crash report.
Build fingerprint: 'xiaomi/mido/mido:7.0/NRD90M/V11.0.2.0.NCFMIXM:user/release-keys'
Revision: '0'
ABI: 'arm64'
pid: 25854, tid: 25898, name: pool-1-thread-1 >>> com.example.open.diffusion <<<
signal 4 (SIGILL), code 1 (ILL_ILLOPC), fault addr 0x7f608bb7fc
x0 0000007f60b811e8 x1 000000000000004d x2 0000000000000300 x3 0000000000000300
x4 0000007f3957ce80 x5 0000000000000001 x6 0000000000000001 x7 0000007f5ed6bbc0
x8 0000000000000000 x9 0000007f60b7b000 x10 0000007f60b7b000 x11 0000007f60b7b000
x12 0000007f60b7b000 x13 0000000000000000 x14 0000007f3957cef8 x15 0000000000000000
x16 0000007f60b7ccf8 x17 0000007f8071ae9c x18 0000007f7ba4c990 x19 0000007f5ed6bbc0
x20 0000000000000008 x21 0000000000000300 x22 0000000000000300 x23 000000000000004d
x24 0000007f17d90300 x25 000000000000006f x26 0000000000000001 x27 0000007f5ceed4e8
x28 0000007f60b81248 x29 0000007f5cee9f40 x30 0000007f608bd76c
sp 0000007f5cee9ef0 pc 0000007f608bb7fc pstate 0000000080000000
backtrace:
#00 pc 0000000000b107fc /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#1 pc 0000000000b12768 /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#2 pc 000000000064b174 /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#3 pc 0000000000a16984 /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#4 pc 0000000000a16060 /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#5 pc 0000000000a2d690 /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#6 pc 0000000000a19588 /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#7 pc 0000000000a18e14 /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#8 pc 0000000000a41850 /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#9 pc 0000000000a41460 /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#10 pc 0000000000a42970 /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#11 pc 00000000003d420c /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#12 pc 00000000003d5f20 /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#13 pc 00000000003a8acc /data/app/com.example.open.diffusion-1/base.apk (offset 0x206000)
#14 pc 000000000000d9fc /data/app/com.example.open.diffusion-1/base.apk (offset 0xfd9000)
#15 pc 000000000056cc04 /data/app/com.example.open.diffusion-1/oat/arm64/base.odex (offset 0x56a000)
Hi developer thanks for making this awesome app, i recently got to know about stable diffusion so I was curious to run it locally, I've found that it can work on pc, mac, ios but not android. I finally found it after all.
Can we use other models if so how, for example I want to use cyber realistic model.
hi I have converted 6-7 models , actually I created a colab notebooks which does it automatically . Most of the model works and give some what good result but dreamlike photo real 2 doesn't work . Also how can we load models from storage rather than building it in the app itself? The feature to use different sampler and save images . Let me know if u want the models .
If you can guide me a bit I'll try to modifying the app file so loading from external storage works .
Could you please add save output to gallery function.
Why does the same seed and prompt give a totally different output?
Why does some models crashes when set to 448x448 and 512X512?
你好,我尝试新建一个项目复现这个工程,就直接把代码copy到新项目里。
发现在运行的时候,点击 生成 按钮,就报错:
java.lang.UnsatisfiedLinkError: No implementation found for long ai.onnxruntime.OrtSession$SessionOptions.createOptions(long) (tried Java_ai_onnxruntime_OrtSession_00024SessionOptions_createOptions and Java_ai_onnxruntime_OrtSession_00024SessionOptions_createOptions__J) at ai.onnxruntime.OrtSession$SessionOptions.createOptions(Native Method) at ai.onnxruntime.OrtSession$SessionOptions.<init>(OrtSession.java:530) at com.xiaoguo.demo.diffusion.tokenizer.EngTokenizer.init(EngTokenizer.java:59)
请问在编译过程中,有什么需要注意的么?
新手小白,刚接触这些,不会用。。。。请帮帮我
你好,请问安装好的应用界面里分别是什么?我理解dropdown 192-512应该是生成图像大小,7.5和8分别是什么?
Hi , can the binary be used in linux based pc? maybe in python as pip have onnxruntime . A little guidance will be appreciated
Thank you
Hi
How does this compare to :
https://github.com/ShiftHackZ/Stable-Diffusion-Android
in terms of performance ?
At least you can choose another model here easily (except the convert part), no hardcoded link to get the default model in the app
您好,博主,生成的图片都比较模糊,细节不清晰,生成的人物头像脸部变现,是提示词没写好导致的么,,
Hello, I am very interested in the Java code under the com.example.open.diffusion.core
package (such as VaeDoder.java
, UNet.java
, etc.). Is there any official documentation for this part of the code that can be used as a reference?
If I want to add a new Scheduler
, is there anything I can use as a reference?
您好,现在的apk中的ort模型,是通过pytorch还是onnx量化和转换而来的?能合入原始模型的路径、量化和转ort脚本不? 方便扩展和新增APP功能
我参考了这个链接可以量化fp16的模型,但是替换model生成图像报错error。
Accelerating Stable Diffusion Inference with ONNX Runtime
请问有stable diffusion模型onnx int8量化参考链接或者代码嘛?
你好,现在的apk只支持txt2img,我想尝试添加img2img的功能。我理解的是img2img会把random seed的latent image改为输入图,所以是只需要修改UNet的输入就行了吗?这个部分的话在你写的Java代码里是需要在哪部分修改呢?
How can other sampler be used Like dpm . Using dpm result in better output .
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.