Slide 1
Slide 1 text
Delightful on-device
AI experiences
@POLPIELLADEV @[email protected] 🦸 SWIFT HEROES 2024
Slide 2
Slide 2 text
App maker and content creator based in Barcelona
Hi! I’m Pol
Slide 3
Slide 3 text
No content
Slide 4
Slide 4 text
Get early access to Helm!
Slide 5
Slide 5 text
No content
Slide 6
Slide 6 text
No content
Slide 7
Slide 7 text
No content
Slide 8
Slide 8 text
No content
Slide 9
Slide 9 text
So… I had an idea
Slide 10
Slide 10 text
No content
Slide 11
Slide 11 text
🔐
PRIVACY
FRIENDLY
🎨
HIGHLY
CUSTOMIZABLE
🛫
OFFLINE
FIRST
Slide 12
Slide 12 text
No content
Slide 13
Slide 13 text
I need this NOW!
Slide 14
Slide 14 text
The concepts
Slide 15
Slide 15 text
No content
Slide 16
Slide 16 text
Stable Di
ff
usion
Slide 17
Slide 17 text
No content
Slide 18
Slide 18 text
A neon drawing of an astronaut
fl
oating in space, digital art by Mór
Than, unsplash contest winner, space
art, sci-
fi
, retrowave, synthwave
Slide 19
Slide 19 text
No content
Slide 20
Slide 20 text
https://inpainter.vercel.app/
Slide 21
Slide 21 text
Stable Di
ff
usion
Slide 22
Slide 22 text
Stable Di
ff
usion
+
ControlNet
Slide 23
Slide 23 text
A high quality photo of a
sur
fi
ng dog
Slide 24
Slide 24 text
ControlNet Input
Slide 25
Slide 25 text
ControlNet Input
https://github.com/apple/ml-stable-di
ff
usion?tab=readme-ov-
fi
le#-using-controlnet
Slide 26
Slide 26 text
A drawing of Darth Vader
from Star Wars
Slide 27
Slide 27 text
ControlNet Input
Slide 28
Slide 28 text
ControlNet Input
Slide 29
Slide 29 text
Finding the right
solution
Slide 30
Slide 30 text
No content
Slide 31
Slide 31 text
No content
Slide 32
Slide 32 text
Houston, we have
TWO problems
Slide 33
Slide 33 text
🔐
PRIVACY
FRIENDLY
🎨
HIGHLY
CUSTOMIZABLE
🛫
OFFLINE
FIRST
#1
Slide 34
Slide 34 text
🛫
OFFLINE
FIRST
#1
Slide 35
Slide 35 text
#2
Slide 36
Slide 36 text
Change the
pricing model?
#2
🤑
Slide 37
Slide 37 text
On-device
Stable Di
ff
usion
Slide 38
Slide 38 text
No content
Slide 39
Slide 39 text
⚙
PyTorch to CoreML
Converter
📦
Swi
ft
Package
to load models
Slide 40
Slide 40 text
No content
Slide 41
Slide 41 text
No content
Slide 42
Slide 42 text
No content
Slide 43
Slide 43 text
No content
Slide 44
Slide 44 text
Did somebody say
ControlNet?
Slide 45
Slide 45 text
No content
Slide 46
Slide 46 text
No content
Slide 47
Slide 47 text
No content
Slide 48
Slide 48 text
git clone https:
/ /
github.com/apple/ml-stable-diffusion.git
cd ml-stable-diffusion
Slide 49
Slide 49 text
git clone https:
/ /
github.com/apple/ml-stable-diffusion.git
cd ml-stable-diffusion
python3 -m venv venv
source venv/bin/activate
pip install -e .
Slide 50
Slide 50 text
No content
Slide 51
Slide 51 text
git clone https:
/ /
github.com/apple/ml-stable-diffusion.git
cd ml-stable-diffusion
python3 -m venv venv
source venv/bin/activate
pip install -e .
python -m python_coreml_stable_diffusion.torch2coreml \
--
bundle-resources-for-swift-cli \
--
attention-implementation SPLIT_EINSUM_V2 \
--
convert-unet \
--
convert-text-encoder \
--
convert-vae-decoder \
--
convert-vae-encoder \
--
model-version runwayml/stable-diffusion-v1-5 \
--
unet-support-controlnet \
--
convert-controlnet DionTimmer/controlnet_qrcode-control_v1p_sd15 \
-o generated
Slide 52
Slide 52 text
git clone https:
/ /
github.com/apple/ml-stable-diffusion.git
cd ml-stable-diffusion
python3 -m venv venv
source venv/bin/activate
pip install -e .
python -m python_coreml_stable_diffusion.torch2coreml \
--
bundle-resources-for-swift-cli \
--
attention-implementation SPLIT_EINSUM_V2 \
--
convert-unet \
--
convert-text-encoder \
--
convert-vae-decoder \
--
convert-vae-encoder \
--
model-version runwayml/stable-diffusion-v1-5 \
--
unet-support-controlnet \
--
convert-controlnet DionTimmer/controlnet_qrcode-control_v1p_sd15 \
-o generated
Slide 53
Slide 53 text
git clone https:
/ /
github.com/apple/ml-stable-diffusion.git
cd ml-stable-diffusion
python3 -m venv venv
source venv/bin/activate
pip install -e .
python -m python_coreml_stable_diffusion.torch2coreml \
--
bundle-resources-for-swift-cli \
--
attention-implementation SPLIT_EINSUM_V2 \
--
convert-unet \
--
convert-text-encoder \
--
convert-vae-decoder \
--
convert-vae-encoder \
--
model-version runwayml/stable-diffusion-v1-5 \
--
unet-support-controlnet \
--
convert-controlnet DionTimmer/controlnet_qrcode-control_v1p_sd15 \
-o generated
Slide 54
Slide 54 text
git clone https:
/ /
github.com/apple/ml-stable-diffusion.git
cd ml-stable-diffusion
python3 -m venv venv
source venv/bin/activate
pip install -e .
python -m python_coreml_stable_diffusion.torch2coreml \
--
bundle-resources-for-swift-cli \
--
attention-implementation SPLIT_EINSUM_V2 \
--
convert-unet \
--
convert-text-encoder \
--
convert-vae-decoder \
--
convert-vae-encoder \
--
model-version runwayml/stable-diffusion-v1-5 \
--
unet-support-controlnet \
--
convert-controlnet DionTimmer/controlnet_qrcode-control_v1p_sd15 \
-o generated
Slide 55
Slide 55 text
No content
Slide 56
Slide 56 text
Loading Stable
Di
ff
usion models
Slide 57
Slide 57 text
//
swift-tools-version: 5.9
import PackageDescription
let package = Package(
name: "StableDiffusionControlNet",
platforms: [.macOS(.v14)],
dependencies: [
],
targets: [
]
)
Slide 58
Slide 58 text
//
swift-tools-version: 5.9
import PackageDescription
let package = Package(
name: "StableDiffusionControlNet",
platforms: [.macOS(.v14)],
dependencies: [
],
targets: [
]
)
.package(url: "https:
//
github.com/apple/ml-stable-diffusion.git", exact: "1.1.0")
Slide 59
Slide 59 text
//
swift-tools-version: 5.9
import PackageDescription
let package = Package(
name: "StableDiffusionControlNet",
platforms: [.macOS(.v14)],
dependencies: [
.package(url: "https:
//
github.com/apple/ml-stable-diffusion.git", exact: "1.1.0")
],
targets: [
]
)
.executableTarget(
name: "StableDiffusionControlNet",
dependencies: [.product(name: "StableDiffusion", package: "ml-stable-diffusion")],
resources: [.process("Resources")]
),
Slide 60
Slide 60 text
import Foundation
import StableDiffusion
import CoreML
func generate(prompt: String, startingImageURL: URL, numberOfImages: Int) async throws
->
[CGImage] {
guard let resourcesURL = Bundle.module.url(forResource: "Resources", withExtension: nil)
?.
path() else {
return []
}
let url = URL(fileURLWithPath: resourcesURL)
let configuration = MLModelConfiguration()
configuration.computeUnits = .all
let pipeline = try StableDiffusionPipeline(resourcesAt: url,
controlNet: ["DiontimmerControlnetQrcodeControlV1PSd15"],
configuration: configuration,
disableSafety: false,
reduceMemory: true)
try pipeline.loadResources()
let startingNSImage = NSImage(contentsOf: startingImageURL)
?.
resized(to: .init(width: 512, height: 512))
guard let startingImage = startingNSImage
?.
cgImage(forProposedRect: nil, context: nil, hints: nil) else {
return []
}
var pipelineConfig = StableDiffusionPipeline.Configuration(prompt: prompt)
pipelineConfig.negativePrompt = "ugly, disfigured, low quality, blurry, nsfw"
pipelineConfig.controlNetInputs = [startingImage]
pipelineConfig.startingImage = startingImage
pipelineConfig.useDenoisedIntermediates = true
pipelineConfig.strength = 0.9
pipelineConfig.seed = UInt32.random(in: (0
..
<
UInt32.max))
pipelineConfig.guidanceScale = 7.5
pipelineConfig.stepCount = 50
pipelineConfig.originalSize = 512
pipelineConfig.targetSize = 512
pipelineConfig.imageCount = numberOfImages
return try pipeline.generateImages(configuration: pipelineConfig, progressHandler: { _ in true })
.compactMap { $0 }
}
Slide 61
Slide 61 text
import Foundation
import StableDiffusion
import CoreML
func generate(prompt: String, startingImageURL: URL, numberOfImages: Int) async throws
->
[CGImage] {
guard let resourcesURL = Bundle.module.url(forResource: "Resources", withExtension: nil)
?.
path() else {
return []
}
let url = URL(fileURLWithPath: resourcesURL)
let configuration = MLModelConfiguration()
configuration.computeUnits = .all
let pipeline = try StableDiffusionPipeline(resourcesAt: url,
controlNet: ["DiontimmerControlnetQrcodeControlV1PSd15"],
configuration: configuration,
disableSafety: false,
reduceMemory: true)
try pipeline.loadResources()
let startingNSImage = NSImage(contentsOf: startingImageURL)
?.
resized(to: .init(width: 512, height: 512))
guard let startingImage = startingNSImage
?.
cgImage(forProposedRect: nil, context: nil, hints: nil) else {
return []
}
var pipelineConfig = StableDiffusionPipeline.Configuration(prompt: prompt)
pipelineConfig.negativePrompt = "ugly, disfigured, low quality, blurry, nsfw"
pipelineConfig.controlNetInputs = [startingImage]
pipelineConfig.startingImage = startingImage
pipelineConfig.useDenoisedIntermediates = true
pipelineConfig.strength = 0.9
pipelineConfig.seed = UInt32.random(in: (0
..
<
UInt32.max))
pipelineConfig.guidanceScale = 7.5
pipelineConfig.stepCount = 50
pipelineConfig.originalSize = 512
pipelineConfig.targetSize = 512
pipelineConfig.imageCount = numberOfImages
return try pipeline.generateImages(configuration: pipelineConfig, progressHandler: { _ in true })
.compactMap { $0 }
}
Slide 62
Slide 62 text
import Foundation
import StableDiffusion
import CoreML
func generate(prompt: String, startingImageURL: URL, numberOfImages: Int) async throws
->
[CGImage] {
guard let resourcesURL = Bundle.module.url(forResource: "Resources", withExtension: nil)
?.
path() else {
return []
}
let url = URL(fileURLWithPath: resourcesURL)
let configuration = MLModelConfiguration()
configuration.computeUnits = .all
let pipeline = try StableDiffusionPipeline(resourcesAt: url,
controlNet: ["DiontimmerControlnetQrcodeControlV1PSd15"],
configuration: configuration,
disableSafety: false,
reduceMemory: true)
try pipeline.loadResources()
let startingNSImage = NSImage(contentsOf: startingImageURL)
?.
resized(to: .init(width: 512, height: 512))
guard let startingImage = startingNSImage
?.
cgImage(forProposedRect: nil, context: nil, hints: nil) else {
return []
}
var pipelineConfig = StableDiffusionPipeline.Configuration(prompt: prompt)
pipelineConfig.negativePrompt = "ugly, disfigured, low quality, blurry, nsfw"
pipelineConfig.controlNetInputs = [startingImage]
pipelineConfig.startingImage = startingImage
pipelineConfig.useDenoisedIntermediates = true
pipelineConfig.strength = 0.9
pipelineConfig.seed = UInt32.random(in: (0
..
<
UInt32.max))
pipelineConfig.guidanceScale = 7.5
pipelineConfig.stepCount = 50
pipelineConfig.originalSize = 512
pipelineConfig.targetSize = 512
pipelineConfig.imageCount = numberOfImages
return try pipeline.generateImages(configuration: pipelineConfig, progressHandler: { _ in true })
.compactMap { $0 }
}
Slide 63
Slide 63 text
import Foundation
import StableDiffusion
import CoreML
func generate(prompt: String, startingImageURL: URL, numberOfImages: Int) async throws
->
[CGImage] {
guard let resourcesURL = Bundle.module.url(forResource: "Resources", withExtension: nil)
?.
path() else {
return []
}
let url = URL(fileURLWithPath: resourcesURL)
let configuration = MLModelConfiguration()
configuration.computeUnits = .all
let pipeline = try StableDiffusionPipeline(resourcesAt: url,
controlNet: ["DiontimmerControlnetQrcodeControlV1PSd15"],
configuration: configuration,
disableSafety: false,
reduceMemory: true)
try pipeline.loadResources()
let startingNSImage = NSImage(contentsOf: startingImageURL)
?.
resized(to: .init(width: 512, height: 512))
guard let startingImage = startingNSImage
?.
cgImage(forProposedRect: nil, context: nil, hints: nil) else {
return []
}
var pipelineConfig = StableDiffusionPipeline.Configuration(prompt: prompt)
pipelineConfig.negativePrompt = "ugly, disfigured, low quality, blurry, nsfw"
pipelineConfig.controlNetInputs = [startingImage]
pipelineConfig.startingImage = startingImage
pipelineConfig.useDenoisedIntermediates = true
pipelineConfig.strength = 0.9
pipelineConfig.seed = UInt32.random(in: (0
..
<
UInt32.max))
pipelineConfig.guidanceScale = 7.5
pipelineConfig.stepCount = 50
pipelineConfig.originalSize = 512
pipelineConfig.targetSize = 512
pipelineConfig.imageCount = numberOfImages
return try pipeline.generateImages(configuration: pipelineConfig, progressHandler: { _ in true })
.compactMap { $0 }
}
Slide 64
Slide 64 text
import Foundation
import StableDiffusion
import CoreML
func generate(prompt: String, startingImageURL: URL, numberOfImages: Int) async throws
->
[CGImage] {
guard let resourcesURL = Bundle.module.url(forResource: "Resources", withExtension: nil)
?.
path() else {
return []
}
let url = URL(fileURLWithPath: resourcesURL)
let configuration = MLModelConfiguration()
configuration.computeUnits = .all
let pipeline = try StableDiffusionPipeline(resourcesAt: url,
controlNet: ["DiontimmerControlnetQrcodeControlV1PSd15"],
configuration: configuration,
disableSafety: false,
reduceMemory: true)
try pipeline.loadResources()
let startingNSImage = NSImage(contentsOf: startingImageURL)
?.
resized(to: .init(width: 512, height: 512))
guard let startingImage = startingNSImage
?.
cgImage(forProposedRect: nil, context: nil, hints: nil) else {
return []
}
var pipelineConfig = StableDiffusionPipeline.Configuration(prompt: prompt)
pipelineConfig.negativePrompt = "ugly, disfigured, low quality, blurry, nsfw"
pipelineConfig.controlNetInputs = [startingImage]
pipelineConfig.startingImage = startingImage
pipelineConfig.useDenoisedIntermediates = true
pipelineConfig.strength = 0.9
pipelineConfig.seed = UInt32.random(in: (0
..
<
UInt32.max))
pipelineConfig.guidanceScale = 7.5
pipelineConfig.stepCount = 50
pipelineConfig.originalSize = 512
pipelineConfig.targetSize = 512
pipelineConfig.imageCount = numberOfImages
return try pipeline.generateImages(configuration: pipelineConfig, progressHandler: { _ in true })
.compactMap { $0 }
}
Slide 65
Slide 65 text
import Foundation
import StableDiffusion
import CoreML
func generate(prompt: String, startingImageURL: URL, numberOfImages: Int) async throws
->
[CGImage] {
guard let resourcesURL = Bundle.module.url(forResource: "Resources", withExtension: nil)
?.
path() else {
return []
}
let url = URL(fileURLWithPath: resourcesURL)
let configuration = MLModelConfiguration()
configuration.computeUnits = .all
let pipeline = try StableDiffusionPipeline(resourcesAt: url,
controlNet: ["DiontimmerControlnetQrcodeControlV1PSd15"],
configuration: configuration,
disableSafety: false,
reduceMemory: true)
try pipeline.loadResources()
let startingNSImage = NSImage(contentsOf: startingImageURL)
?.
resized(to: .init(width: 512, height: 512))
guard let startingImage = startingNSImage
?.
cgImage(forProposedRect: nil, context: nil, hints: nil) else {
return []
}
var pipelineConfig = StableDiffusionPipeline.Configuration(prompt: prompt)
pipelineConfig.negativePrompt = "ugly, disfigured, low quality, blurry, nsfw"
pipelineConfig.controlNetInputs = [startingImage]
pipelineConfig.startingImage = startingImage
pipelineConfig.useDenoisedIntermediates = true
pipelineConfig.strength = 0.9
pipelineConfig.seed = UInt32.random(in: (0
..
<
UInt32.max))
pipelineConfig.guidanceScale = 7.5
pipelineConfig.stepCount = 50
pipelineConfig.originalSize = 512
pipelineConfig.targetSize = 512
pipelineConfig.imageCount = numberOfImages
return try pipeline.generateImages(configuration: pipelineConfig, progressHandler: { _ in true })
.compactMap { $0 }
}
Slide 66
Slide 66 text
import Foundation
import StableDiffusion
import CoreML
func generate(prompt: String, startingImageURL: URL, numberOfImages: Int) async throws
->
[CGImage] {
guard let resourcesURL = Bundle.module.url(forResource: "Resources", withExtension: nil)
?.
path() else {
return []
}
let url = URL(fileURLWithPath: resourcesURL)
let configuration = MLModelConfiguration()
configuration.computeUnits = .all
let pipeline = try StableDiffusionPipeline(resourcesAt: url,
controlNet: ["DiontimmerControlnetQrcodeControlV1PSd15"],
configuration: configuration,
disableSafety: false,
reduceMemory: false)
try pipeline.loadResources()
let startingNSImage = NSImage(contentsOf: startingImageURL)
?.
resized(to: .init(width: 512, height: 512))
guard let startingImage = startingNSImage
?.
cgImage(forProposedRect: nil, context: nil, hints: nil) else {
return []
}
var pipelineConfig = StableDiffusionPipeline.Configuration(prompt: prompt)
pipelineConfig.negativePrompt = "ugly, disfigured, low quality, blurry, nsfw"
pipelineConfig.controlNetInputs = [startingImage]
pipelineConfig.startingImage = startingImage
pipelineConfig.useDenoisedIntermediates = true
pipelineConfig.strength = 0.9
pipelineConfig.seed = UInt32.random(in: (0
..
<
UInt32.max))
pipelineConfig.guidanceScale = 7.5
pipelineConfig.stepCount = 50
pipelineConfig.originalSize = 512
pipelineConfig.targetSize = 512
pipelineConfig.imageCount = numberOfImages
return try pipeline.generateImages(configuration: pipelineConfig, progressHandler: { _ in true })
.compactMap { $0 }
}
Slide 67
Slide 67 text
import Foundation
import StableDiffusion
import CoreML
func generate(prompt: String, startingImageURL: URL, numberOfImages: Int) async throws
->
[CGImage] {
guard let resourcesURL = Bundle.module.url(forResource: "Resources", withExtension: nil)
?.
path() else {
return []
}
let url = URL(fileURLWithPath: resourcesURL)
let configuration = MLModelConfiguration()
configuration.computeUnits = .all
let pipeline = try StableDiffusionPipeline(resourcesAt: url,
controlNet: ["DiontimmerControlnetQrcodeControlV1PSd15"],
configuration: configuration,
disableSafety: false,
reduceMemory: false)
try pipeline.loadResources()
let startingNSImage = NSImage(contentsOf: startingImageURL)
?.
resized(to: .init(width: 512, height: 512))
guard let startingImage = startingNSImage
?.
cgImage(forProposedRect: nil, context: nil, hints: nil) else {
return []
}
var pipelineConfig = StableDiffusionPipeline.Configuration(prompt: prompt)
pipelineConfig.negativePrompt = "ugly, disfigured, low quality, blurry, nsfw"
pipelineConfig.controlNetInputs = [startingImage]
pipelineConfig.startingImage = startingImage
pipelineConfig.useDenoisedIntermediates = true
pipelineConfig.strength = 0.9
pipelineConfig.seed = UInt32.random(in: (0
..
<
UInt32.max))
pipelineConfig.guidanceScale = 7.5
pipelineConfig.stepCount = 50
pipelineConfig.originalSize = 512
pipelineConfig.targetSize = 512
pipelineConfig.imageCount = numberOfImages
return try pipeline.generateImages(configuration: pipelineConfig, progressHandler: { _ in true })
.compactMap { $0 }
}
Slide 68
Slide 68 text
let prompt = """
Style-NebMagic, award winning photo, A Dark-Eyed Junco, sitting Great Basin National Park,
intricate, nature background, wildlife photography, hyper realistic, Style-LostTemple,
deep shadow, high contrast, dark, sunrise, morning, full moon
"""
Slide 69
Slide 69 text
let prompt = """
Style-NebMagic, award winning photo, A Dark-Eyed Junco, sitting Great Basin National Park,
intricate, nature background, wildlife photography, hyper realistic, Style-LostTemple,
deep shadow, high contrast, dark, sunrise, morning, full moon
"""
let url = URL(filePath: "/my-qr-code.png")
Slide 70
Slide 70 text
let prompt = """
Style-NebMagic, award winning photo, A Dark-Eyed Junco, sitting Great Basin National Park,
intricate, nature background, wildlife photography, hyper realistic, Style-LostTemple,
deep shadow, high contrast, dark, sunrise, morning, full moon
"""
let url = URL(filePath: "/my-qr-code.png")
let image = try await generate(
prompt: prompt,
startingImageURL: url,
numberOfImages: 1
)
Slide 71
Slide 71 text
No content
Slide 72
Slide 72 text
https://github.com/huggingface/swi
ft
-coreml-di
ff
users
Slide 73
Slide 73 text
One more thing…
Slide 74
Slide 74 text
No content
Slide 75
Slide 75 text
No content
Slide 76
Slide 76 text
Host your models
remotely
And load them on demand
Slide 77
Slide 77 text
No content
Slide 78
Slide 78 text
let modelURL = URL(string: "https:
//
huggingface.co/:user/:model/resolve/main/:file.zip?download=true")!
let (location, downloadFileResponse) = try await URLSession.shared.download(from: modelURL)
guard let httpResponse = downloadFileResponse as? HTTPURLResponse,
httpResponse.statusCode
==
200 else {
exit(1)
}
try FileManager.default
.moveItem(
at: location,
to: URL.desktopDirectory.appending(component: "model.zip")
)
Model from HuggingFace
Slide 79
Slide 79 text
Thank you for listening!
@POLPIELLADEV @[email protected] 🦸 SWIFT HEROES 2024