🎨 Data Designer Tutorial: Image-to-Image Editing¶

📚 What you'll learn¶

This notebook shows how to chain image generation columns: first generate animal portraits from text, then edit those generated images by adding accessories and changing styles—all without loading external datasets.

• 🖼️ Text-to-image generation: Generate images from text prompts
• 🔗 Chaining image columns: Use ImageContext to pass generated images to a follow-up editing column
• 🎲 Sampler-driven diversity: Combine sampled accessories and settings for varied edits

This tutorial uses an autoregressive model (one that supports both text-to-image and image-to-image generation via the chat completions API). Diffusion models (DALL·E, Stable Diffusion, etc.) do not support image context—see Tutorial 5 for text-to-image generation with diffusion models.

Prerequisites: This tutorial uses OpenRouter with the Flux 2 Pro model. Set OPENROUTER_API_KEY in your environment before running.

If this is your first time using Data Designer, we recommend starting with the first notebook in this tutorial series.

📦 Import Data Designer¶

• data_designer.config provides the configuration API.
• DataDesigner is the main interface for generation.

In [1]:

import base64
from pathlib import Path

from IPython.display import Image as IPImage
from IPython.display import display

import data_designer.config as dd
from data_designer.interface import DataDesigner

import base64 from pathlib import Path from IPython.display import Image as IPImage from IPython.display import display import data_designer.config as dd from data_designer.interface import DataDesigner

⚙️ Initialize the Data Designer interface¶

We initialize Data Designer without arguments here—the image model is configured explicitly in the next cell.

In [2]:

data_designer = DataDesigner()

data_designer = DataDesigner()

🎛️ Define an image model¶

We need an autoregressive model that supports both text-to-image and image-to-image generation via the chat completions API. This lets us generate images from text and then pass those images as context for editing.

• Use ImageInferenceParams so Data Designer treats this model as an image generator.
• Image-specific options are model-dependent; pass them via extra_body.

Note: This tutorial uses the Flux 2 Pro model via OpenRouter. Set OPENROUTER_API_KEY in your environment.

In [3]:

MODEL_PROVIDER = "openrouter"
MODEL_ID = "black-forest-labs/flux.2-pro"
MODEL_ALIAS = "image-model"

model_configs = [
    dd.ModelConfig(
        alias=MODEL_ALIAS,
        model=MODEL_ID,
        provider=MODEL_PROVIDER,
        inference_parameters=dd.ImageInferenceParams(
            extra_body={"height": 512, "width": 512},
        ),
    )
]

MODEL_PROVIDER = "openrouter" MODEL_ID = "black-forest-labs/flux.2-pro" MODEL_ALIAS = "image-model" model_configs = [ dd.ModelConfig( alias=MODEL_ALIAS, model=MODEL_ID, provider=MODEL_PROVIDER, inference_parameters=dd.ImageInferenceParams( extra_body={"height": 512, "width": 512}, ), ) ]

🏗️ Build the configuration¶

We chain two image generation columns:

1. Sampler columns — randomly sample animal types, accessories, settings, and art styles
2. First image column — generate an animal portrait from a text prompt
3. Second image column with context — edit the generated portrait using ImageContext

In [4]:

config_builder = dd.DataDesignerConfigBuilder(model_configs=model_configs)

# 1. Sampler columns for diversity
config_builder.add_column(
    dd.SamplerColumnConfig(
        name="animal",
        sampler_type=dd.SamplerType.CATEGORY,
        params=dd.CategorySamplerParams(
            values=["cat", "dog", "fox", "owl", "rabbit", "panda"],
        ),
    )
)

config_builder.add_column(
    dd.SamplerColumnConfig(
        name="accessory",
        sampler_type=dd.SamplerType.CATEGORY,
        params=dd.CategorySamplerParams(
            values=[
                "a tiny top hat",
                "oversized sunglasses",
                "a red bow tie",
                "a knitted beanie",
                "a flower crown",
                "a monocle and mustache",
                "a pirate hat and eye patch",
                "a chef hat",
            ],
        ),
    )
)

config_builder.add_column(
    dd.SamplerColumnConfig(
        name="setting",
        sampler_type=dd.SamplerType.CATEGORY,
        params=dd.CategorySamplerParams(
            values=[
                "a cozy living room",
                "a sunny park",
                "a photo studio with soft lighting",
                "a red carpet event",
                "a holiday card backdrop with snowflakes",
                "a tropical beach at sunset",
            ],
        ),
    )
)

config_builder.add_column(
    dd.SamplerColumnConfig(
        name="art_style",
        sampler_type=dd.SamplerType.CATEGORY,
        params=dd.CategorySamplerParams(
            values=[
                "a photorealistic style",
                "a Disney Pixar 3D render",
                "a watercolor painting",
                "a pop art poster",
            ],
        ),
    )
)

# 2. Generate animal portrait from text
config_builder.add_column(
    dd.ImageColumnConfig(
        name="animal_portrait",
        prompt="A close-up portrait photograph of a {{ animal }} looking at the camera, studio lighting, high quality.",
        model_alias=MODEL_ALIAS,
    )
)

# 3. Edit the generated portrait
config_builder.add_column(
    dd.ImageColumnConfig(
        name="edited_portrait",
        prompt=(
            "Edit this {{ animal }} portrait photo. "
            "Add {{ accessory }} on the animal. "
            "Place the {{ animal }} in {{ setting }}. "
            "Render the result in {{ art_style }}. "
            "Keep the animal's face, expression, and features faithful to the original photo."
        ),
        model_alias=MODEL_ALIAS,
        multi_modal_context=[dd.ImageContext(column_name="animal_portrait")],
    )
)

data_designer.validate(config_builder)

config_builder = dd.DataDesignerConfigBuilder(model_configs=model_configs) # 1. Sampler columns for diversity config_builder.add_column( dd.SamplerColumnConfig( name="animal", sampler_type=dd.SamplerType.CATEGORY, params=dd.CategorySamplerParams( values=["cat", "dog", "fox", "owl", "rabbit", "panda"], ), ) ) config_builder.add_column( dd.SamplerColumnConfig( name="accessory", sampler_type=dd.SamplerType.CATEGORY, params=dd.CategorySamplerParams( values=[ "a tiny top hat", "oversized sunglasses", "a red bow tie", "a knitted beanie", "a flower crown", "a monocle and mustache", "a pirate hat and eye patch", "a chef hat", ], ), ) ) config_builder.add_column( dd.SamplerColumnConfig( name="setting", sampler_type=dd.SamplerType.CATEGORY, params=dd.CategorySamplerParams( values=[ "a cozy living room", "a sunny park", "a photo studio with soft lighting", "a red carpet event", "a holiday card backdrop with snowflakes", "a tropical beach at sunset", ], ), ) ) config_builder.add_column( dd.SamplerColumnConfig( name="art_style", sampler_type=dd.SamplerType.CATEGORY, params=dd.CategorySamplerParams( values=[ "a photorealistic style", "a Disney Pixar 3D render", "a watercolor painting", "a pop art poster", ], ), ) ) # 2. Generate animal portrait from text config_builder.add_column( dd.ImageColumnConfig( name="animal_portrait", prompt="A close-up portrait photograph of a {{ animal }} looking at the camera, studio lighting, high quality.", model_alias=MODEL_ALIAS, ) ) # 3. Edit the generated portrait config_builder.add_column( dd.ImageColumnConfig( name="edited_portrait", prompt=( "Edit this {{ animal }} portrait photo. " "Add {{ accessory }} on the animal. " "Place the {{ animal }} in {{ setting }}. " "Render the result in {{ art_style }}. " "Keep the animal's face, expression, and features faithful to the original photo." ), model_alias=MODEL_ALIAS, multi_modal_context=[dd.ImageContext(column_name="animal_portrait")], ) ) data_designer.validate(config_builder)

[12:19:07] [INFO] ✅ Validation passed

🔁 Preview: quick iteration¶

In preview mode, generated images are stored as base64 strings in the dataframe. Use this to iterate on your prompts, accessories, and sampler values before scaling up.

In [5]:

preview = data_designer.preview(config_builder, num_records=2)

preview = data_designer.preview(config_builder, num_records=2)

[12:19:07] [INFO] 👁️ Preview generation in progress

[12:19:07] [INFO] ✅ Validation passed

[12:19:08] [INFO] ⛓️ Sorting column configs into a Directed Acyclic Graph

[12:19:08] [INFO] 🩺 Running health checks for models...

[12:19:08] [INFO]   |-- 👀 Checking 'black-forest-labs/flux.2-pro' in provider named 'openrouter' for model alias 'image-model'...

[12:19:17] [INFO]   |-- ✅ Passed!

[12:19:17] [INFO] 🎲 Preparing samplers to generate 2 records across 4 columns

[12:19:17] [INFO] 🖼️ image model config for column 'animal_portrait'

[12:19:17] [INFO]   |-- model: 'black-forest-labs/flux.2-pro'

[12:19:17] [INFO]   |-- model alias: 'image-model'

[12:19:17] [INFO]   |-- model provider: 'openrouter'

[12:19:17] [INFO]   |-- inference parameters:

[12:19:17] [INFO]   |  |-- generation_type=image

[12:19:17] [INFO]   |  |-- max_parallel_requests=4

[12:19:17] [INFO]   |  |-- extra_body={'height': 512, 'width': 512}

[12:19:17] [INFO] ⚡️ Processing image column 'animal_portrait' with 4 concurrent workers

[12:19:17] [INFO] ⏱️ image column 'animal_portrait' will report progress after each record

[12:19:25] [INFO]   |-- 😐 image column 'animal_portrait' progress: 1/2 (50%) complete, 1 ok, 0 failed, 0.13 rec/s, eta 7.9s

[12:19:27] [INFO]   |-- 🤩 image column 'animal_portrait' progress: 2/2 (100%) complete, 2 ok, 0 failed, 0.20 rec/s, eta 0.0s

[12:19:27] [INFO] 🖼️ image model config for column 'edited_portrait'

[12:19:27] [INFO]   |-- model: 'black-forest-labs/flux.2-pro'

[12:19:27] [INFO]   |-- model alias: 'image-model'

[12:19:27] [INFO]   |-- model provider: 'openrouter'

[12:19:27] [INFO]   |-- inference parameters:

[12:19:27] [INFO]   |  |-- generation_type=image

[12:19:27] [INFO]   |  |-- max_parallel_requests=4

[12:19:27] [INFO]   |  |-- extra_body={'height': 512, 'width': 512}

[12:19:27] [INFO] ⚡️ Processing image column 'edited_portrait' with 4 concurrent workers

[12:19:27] [INFO] ⏱️ image column 'edited_portrait' will report progress after each record

[12:19:41] [INFO]   |-- ⛅ image column 'edited_portrait' progress: 1/2 (50%) complete, 1 ok, 0 failed, 0.08 rec/s, eta 13.3s

[12:19:42] [INFO]   |-- ☀️ image column 'edited_portrait' progress: 2/2 (100%) complete, 2 ok, 0 failed, 0.14 rec/s, eta 0.0s

[12:19:42] [INFO] 📊 Model usage summary:

[12:19:42] [INFO]   |-- model: black-forest-labs/flux.2-pro

[12:19:42] [INFO]   |-- tokens: input=0, output=0, total=0, tps=0

[12:19:42] [INFO]   |-- requests: success=4, failed=0, total=4, rpm=9

[12:19:42] [INFO]   |-- images: total=4

[12:19:42] [INFO] 📐 Measuring dataset column statistics:

[12:19:42] [INFO]   |-- 🎲 column: 'animal'

[12:19:42] [INFO]   |-- 🎲 column: 'accessory'

[12:19:42] [INFO]   |-- 🎲 column: 'setting'

[12:19:42] [INFO]   |-- 🎲 column: 'art_style'

[12:19:42] [INFO]   |-- 🖼️ column: 'animal_portrait'

[12:19:42] [INFO]   |-- 🖼️ column: 'edited_portrait'

[12:19:42] [INFO] 🎆 Preview complete!

In [6]:

for i in range(len(preview.dataset)):
    preview.display_sample_record()

for i in range(len(preview.dataset)): preview.display_sample_record()

[index: 0]

                                                                                                              
                                              Generated Columns                                               
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Name                                 ┃ Value                                                               ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ animal                               │ owl                                                                 │
├──────────────────────────────────────┼─────────────────────────────────────────────────────────────────────┤
│ accessory                            │ a tiny top hat                                                      │
├──────────────────────────────────────┼─────────────────────────────────────────────────────────────────────┤
│ setting                              │ a red carpet event                                                  │
├──────────────────────────────────────┼─────────────────────────────────────────────────────────────────────┤
│ art_style                            │ a pop art poster                                                    │
└──────────────────────────────────────┴─────────────────────────────────────────────────────────────────────┘
                                                                                                              
                                                                                                              
                                                    Images                                                    
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Name                                   ┃ Preview                                                           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ animal_portrait                        │ [0] <base64, 1713836 chars>                                       │
├────────────────────────────────────────┼───────────────────────────────────────────────────────────────────┤
│ edited_portrait                        │ [0] <base64, 2175740 chars>                                       │
└────────────────────────────────────────┴───────────────────────────────────────────────────────────────────┘
                                                                                                              

🖼️ animal_portrait[0]

🖼️ edited_portrait[0]

[index: 1]

                                                                                                              
                                              Generated Columns                                               
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Name                           ┃ Value                                                                     ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ animal                         │ fox                                                                       │
├────────────────────────────────┼───────────────────────────────────────────────────────────────────────────┤
│ accessory                      │ a tiny top hat                                                            │
├────────────────────────────────┼───────────────────────────────────────────────────────────────────────────┤
│ setting                        │ a red carpet event                                                        │
├────────────────────────────────┼───────────────────────────────────────────────────────────────────────────┤
│ art_style                      │ a Disney Pixar 3D render                                                  │
└────────────────────────────────┴───────────────────────────────────────────────────────────────────────────┘
                                                                                                              
                                                                                                              
                                                    Images                                                    
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Name                                   ┃ Preview                                                           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ animal_portrait                        │ [0] <base64, 1822088 chars>                                       │
├────────────────────────────────────────┼───────────────────────────────────────────────────────────────────┤
│ edited_portrait                        │ [0] <base64, 1800368 chars>                                       │
└────────────────────────────────────────┴───────────────────────────────────────────────────────────────────┘
                                                                                                              

🖼️ animal_portrait[0]

🖼️ edited_portrait[0]

In [7]:

preview.dataset

preview.dataset

Out[7]:

	animal	accessory	setting	art_style	animal_portrait	edited_portrait
0	owl	a tiny top hat	a red carpet event	a pop art poster	[iVBORw0KGgoAAAANSUhEUgAABAAAAAMACAIAAAA12IJaA...	[iVBORw0KGgoAAAANSUhEUgAABAAAAAMACAIAAAA12IJaA...
1	fox	a tiny top hat	a red carpet event	a Disney Pixar 3D render	[iVBORw0KGgoAAAANSUhEUgAABAAAAAMACAIAAAA12IJaA...	[iVBORw0KGgoAAAANSUhEUgAABAAAAAMACAIAAAA12IJaA...

🔎 Compare original vs edited¶

Let's display the generated animal portraits next to their edited versions.

In [8]:

def display_image(image_value, base_path: Path | None = None) -> None:
    """Display an image from base64 (preview mode) or file path (create mode)."""
    values = [image_value] if isinstance(image_value, str) else list(image_value)
    for value in values:
        if base_path is not None:
            display(IPImage(filename=str(base_path / value)))
        else:
            display(IPImage(data=base64.b64decode(value)))


def display_before_after(row, index: int, base_path: Path | None = None) -> None:
    """Display original portrait vs edited version for a single record."""
    print(f"\n{'=' * 60}")
    print(f"Record {index}: {row['animal']} wearing {row['accessory']}")
    print(f"Setting: {row['setting']}, Style: {row['art_style']}")
    print(f"{'=' * 60}")

    print("\n📷 Generated portrait:")
    display_image(row["animal_portrait"], base_path)

    print("\n🎨 Edited version:")
    display_image(row["edited_portrait"], base_path)

def display_image(image_value, base_path: Path | None = None) -> None: """Display an image from base64 (preview mode) or file path (create mode).""" values = [image_value] if isinstance(image_value, str) else list(image_value) for value in values: if base_path is not None: display(IPImage(filename=str(base_path / value))) else: display(IPImage(data=base64.b64decode(value))) def display_before_after(row, index: int, base_path: Path | None = None) -> None: """Display original portrait vs edited version for a single record.""" print(f"\n{'=' * 60}") print(f"Record {index}: {row['animal']} wearing {row['accessory']}") print(f"Setting: {row['setting']}, Style: {row['art_style']}") print(f"{'=' * 60}") print("\n📷 Generated portrait:") display_image(row["animal_portrait"], base_path) print("\n🎨 Edited version:") display_image(row["edited_portrait"], base_path)

In [9]:

for index, row in preview.dataset.iterrows():
    display_before_after(row, index)

for index, row in preview.dataset.iterrows(): display_before_after(row, index)

============================================================
Record 0: owl wearing a tiny top hat
Setting: a red carpet event, Style: a pop art poster
============================================================

📷 Generated portrait:

🎨 Edited version:

============================================================
Record 1: fox wearing a tiny top hat
Setting: a red carpet event, Style: a Disney Pixar 3D render
============================================================

📷 Generated portrait:

🎨 Edited version:

🆙 Create at scale¶

In create mode, images are saved to disk in images/<column_name>/ folders with UUID filenames. The dataframe stores relative paths. ImageContext auto-detection handles this transparently—generated file paths are resolved to base64 before being sent to the model for editing.

In [10]:

results = data_designer.create(config_builder, num_records=5, dataset_name="tutorial-6-edited-images")

results = data_designer.create(config_builder, num_records=5, dataset_name="tutorial-6-edited-images")

[12:19:42] [INFO] 🎨 Creating Data Designer dataset

[12:19:42] [INFO] ✅ Validation passed

[12:19:42] [INFO] ⛓️ Sorting column configs into a Directed Acyclic Graph

[12:19:42] [INFO] 🩺 Running health checks for models...

[12:19:42] [INFO]   |-- 👀 Checking 'black-forest-labs/flux.2-pro' in provider named 'openrouter' for model alias 'image-model'...

[12:19:51] [INFO]   |-- ✅ Passed!

[12:19:51] [INFO] ⏳ Processing batch 1 of 1

[12:19:51] [INFO] 🎲 Preparing samplers to generate 5 records across 4 columns

[12:19:51] [INFO] 🖼️ image model config for column 'animal_portrait'

[12:19:51] [INFO]   |-- model: 'black-forest-labs/flux.2-pro'

[12:19:51] [INFO]   |-- model alias: 'image-model'

[12:19:51] [INFO]   |-- model provider: 'openrouter'

[12:19:51] [INFO]   |-- inference parameters:

[12:19:51] [INFO]   |  |-- generation_type=image

[12:19:51] [INFO]   |  |-- max_parallel_requests=4

[12:19:51] [INFO]   |  |-- extra_body={'height': 512, 'width': 512}

[12:19:51] [INFO] ⚡️ Processing image column 'animal_portrait' with 4 concurrent workers

[12:19:51] [INFO] ⏱️ image column 'animal_portrait' will report progress after each record

[12:20:00] [INFO]   |-- 🥚 image column 'animal_portrait' progress: 1/5 (20%) complete, 1 ok, 0 failed, 0.11 rec/s, eta 35.1s

[12:20:00] [INFO]   |-- 🐣 image column 'animal_portrait' progress: 2/5 (40%) complete, 2 ok, 0 failed, 0.21 rec/s, eta 14.4s

[12:20:00] [INFO]   |-- 🐥 image column 'animal_portrait' progress: 3/5 (60%) complete, 3 ok, 0 failed, 0.31 rec/s, eta 6.4s

[12:20:01] [INFO]   |-- 🐤 image column 'animal_portrait' progress: 4/5 (80%) complete, 4 ok, 0 failed, 0.41 rec/s, eta 2.5s

[12:20:10] [INFO]   |-- 🐔 image column 'animal_portrait' progress: 5/5 (100%) complete, 5 ok, 0 failed, 0.27 rec/s, eta 0.0s

[12:20:10] [INFO] 🖼️ image model config for column 'edited_portrait'

[12:20:10] [INFO]   |-- model: 'black-forest-labs/flux.2-pro'

[12:20:10] [INFO]   |-- model alias: 'image-model'

[12:20:10] [INFO]   |-- model provider: 'openrouter'

[12:20:10] [INFO]   |-- inference parameters:

[12:20:10] [INFO]   |  |-- generation_type=image

[12:20:10] [INFO]   |  |-- max_parallel_requests=4

[12:20:10] [INFO]   |  |-- extra_body={'height': 512, 'width': 512}

[12:20:10] [INFO] ⚡️ Processing image column 'edited_portrait' with 4 concurrent workers

[12:20:10] [INFO] ⏱️ image column 'edited_portrait' will report progress after each record

[12:20:22] [INFO]   |-- 😴 image column 'edited_portrait' progress: 1/5 (20%) complete, 1 ok, 0 failed, 0.08 rec/s, eta 50.2s

[12:20:25] [INFO]   |-- 🥱 image column 'edited_portrait' progress: 2/5 (40%) complete, 2 ok, 0 failed, 0.13 rec/s, eta 23.4s

[12:20:25] [INFO]   |-- 😐 image column 'edited_portrait' progress: 3/5 (60%) complete, 3 ok, 0 failed, 0.19 rec/s, eta 10.6s

[12:20:27] [INFO]   |-- 😊 image column 'edited_portrait' progress: 4/5 (80%) complete, 4 ok, 0 failed, 0.24 rec/s, eta 4.2s

[12:20:45] [INFO]   |-- 🤩 image column 'edited_portrait' progress: 5/5 (100%) complete, 5 ok, 0 failed, 0.14 rec/s, eta 0.0s

[12:20:45] [INFO] 📊 Model usage summary:

[12:20:45] [INFO]   |-- model: black-forest-labs/flux.2-pro

[12:20:45] [INFO]   |-- tokens: input=0, output=0, total=0, tps=0

[12:20:45] [INFO]   |-- requests: success=10, failed=0, total=10, rpm=11

[12:20:45] [INFO]   |-- images: total=10

[12:20:45] [INFO] 📐 Measuring dataset column statistics:

[12:20:45] [INFO]   |-- 🎲 column: 'animal'

[12:20:45] [INFO]   |-- 🎲 column: 'accessory'

[12:20:45] [INFO]   |-- 🎲 column: 'setting'

[12:20:45] [INFO]   |-- 🎲 column: 'art_style'

[12:20:45] [INFO]   |-- 🖼️ column: 'animal_portrait'

[12:20:45] [INFO]   |-- 🖼️ column: 'edited_portrait'

In [11]:

dataset = results.load_dataset()
dataset.head()

dataset = results.load_dataset() dataset.head()

Out[11]:

	animal	accessory	setting	art_style	animal_portrait	edited_portrait
0	owl	a red bow tie	a red carpet event	a watercolor painting	['images/animal_portrait/dff93e50-2905-4774-a3...	['images/edited_portrait/0d9bb828-1a63-4b89-8f...
1	owl	a red bow tie	a sunny park	a photorealistic style	['images/animal_portrait/342ca4af-8fe3-473e-98...	['images/edited_portrait/85846f4f-a41b-4b96-b6...
2	cat	a monocle and mustache	a cozy living room	a watercolor painting	['images/animal_portrait/61aaab70-5da8-4222-b2...	['images/edited_portrait/42f96497-3e8d-42c3-91...
3	cat	a knitted beanie	a photo studio with soft lighting	a Disney Pixar 3D render	['images/animal_portrait/01731a2d-7994-457c-90...	['images/edited_portrait/81f38091-0b45-450e-9a...
4	fox	a red bow tie	a red carpet event	a photorealistic style	['images/animal_portrait/fc983f1f-7158-4f03-bb...	['images/edited_portrait/9443ffe4-bcd2-4191-8a...

In [12]:

for index, row in dataset.head(10).iterrows():
    display_before_after(row, index, base_path=results.artifact_storage.base_dataset_path)

for index, row in dataset.head(10).iterrows(): display_before_after(row, index, base_path=results.artifact_storage.base_dataset_path)

============================================================
Record 0: owl wearing a red bow tie
Setting: a red carpet event, Style: a watercolor painting
============================================================

📷 Generated portrait:

🎨 Edited version:

============================================================
Record 1: owl wearing a red bow tie
Setting: a sunny park, Style: a photorealistic style
============================================================

📷 Generated portrait:

🎨 Edited version:

============================================================
Record 2: cat wearing a monocle and mustache
Setting: a cozy living room, Style: a watercolor painting
============================================================

📷 Generated portrait:

🎨 Edited version:

============================================================
Record 3: cat wearing a knitted beanie
Setting: a photo studio with soft lighting, Style: a Disney Pixar 3D render
============================================================

📷 Generated portrait:

🎨 Edited version:

============================================================
Record 4: fox wearing a red bow tie
Setting: a red carpet event, Style: a photorealistic style
============================================================

📷 Generated portrait:

🎨 Edited version:

⏭️ Next steps¶

• Experiment with different autoregressive models for image generation and editing
• Try more creative editing prompts (style transfer, background replacement, artistic filters)
• Combine image generation with text generation (e.g., generate captions using an LLM-Text column with ImageContext)
• Chain more than two image columns for multi-step editing pipelines

Related tutorials:

• The basics: samplers and LLM text columns
• Providing images as context: image-to-text with VLMs
• Generating images: text-to-image generation with diffusion models