Flux.2 (FLEX2) LoRA Training Complete Tutorial

Virtual product

{{ product.price_format }}
{{ product.origin_price_format }}
-{{ Math.round((1 - product.price / product.origin_price) * 100) }}%
Quantity:
Model: {{ product.model }}
Sold: {{ product.sales_value || 0 }} {{ source.unit_name }}
Purchase Limit: Limit {{ product.order_user_limit }} per user

{{ variable.name }}

{{ value.name }}

{{ bundleConfig.title }}

Editing bundle

Select products to buy together. Choose each variant and quantity.

Quantities are the totals for this purchase. Saving combines the items into one bundle and recalculates the discount.

{{ bundleTierLabel(tier) }}

{{ bundleConfig.name }} Main product

{{ bundlePromotionPreview ? bundleText.after_bundle_discount : bundleText.line_total }}: {{ bundleLinePrice(product) }} {{ bundleLineOriginalPrice(product) }}

Variant and quantity follow your selection above · × {{ quantity }}

{{ bundleStockLabel(product) }}

{{ bundleStockError(product, quantity) }}

{{ bundleLimitError(product, quantity) || bundleLimitLabel(product) }}

{{ item.name }}

{{ bundlePromotionPreview ? bundleText.after_bundle_discount : bundleText.line_total }}: {{ bundleLinePrice(bundleSku(item)) }} {{ bundleLineOriginalPrice(bundleSku(item)) }}
  • {{ sku.label }} {{ bundleStockLabel(sku) }} · {{ bundleStockError(sku, 1) || bundleText.out_of_stock }}
{{ bundleSku(item) ? bundleSku(item).label : bundleText.single_variant }}

{{ bundleStockLabel(bundleSku(item)) }}

{{ bundleStockError(bundleSku(item), item.quantity) }}

Check the selected variants, quantities and stock.

{{ (item.checked && bundleLimitError(bundleSku(item), item.quantity)) || bundleLimitLabel(bundleSku(item)) }}

The main product is excluded. Multiple units of the same product count as one style.

All complementary products have been deselected. Saving will change this to a separate purchase of the main product, without the bundle discount.

{{ bundleSelectedLabel }} Highest eligible discount applied

Unable to calculate. Resolve the message above first.

Subtotal
{{ bundleDisplayMoney('subtotal') }}
Bundle discount
{{ bundlePromotionPreview ? bundleDisplayMoney('discount') : '−' + bundleDisplayMoney('discount') }}
Main product customization fee
{{ bundleDisplayMoney('customization_fee') }}
Currency rounding adjustment
{{ bundleDisplayMoney('rounding_adjustment') }}
{{ bundleDisplayMoney('total') }}

Based on your current selection. Final discounts are confirmed at checkout.

Quantity: {{ quantity * bundleSetQuantity }}

Sets: {{ bundleSetQuantity }}

Customize the main product in your cart or at checkout. Customization fees are charged separately.

Loading…

The ultimate guide to training LoRAs for Flux.2 — the groundbreaking 32B parameter model with native 4 Megapixel (2048x2048) resolution support. Learn to leverage Hex-code color steering and enhanced prompt adherence with this cutting-edge December 2025 training tutorial.

Product Type: Digital Tutorial (HTML + Markdown)

Why Flux.2 for Training?
- 32B parameters: Double the capacity of Flux.1
- Native 4MP resolution: 2048x2048 generation capability
- Hex-code steering: Direct color control via #RRGGBB codes
- Enhanced prompt adherence: Even better natural language understanding
- FP8 quantization support: Train on consumer GPUs with optimization

Tutorial Structure (6 Chapters):

【Chapter 1: Environment & Hardware】
- GPU requirements: RTX 5090 (32GB) optimal, RTX 4090 (24GB) with FP8
- System RAM: 64-128GB for 32B model loading
- Kohya_ss 2025 version with flux2 architecture support
- Model preparation: flux2-dev-fp8.safetensors (mandatory for local training)

【Chapter 2: Dataset Preparation (4MP Era)】
- Resolution recommendations: 1536px-2048px for best quality
- Image quantity: 20-50 high-quality images
- Natural language captioning with JSON-style support
- Trigger word placement and descriptive sentence structure

【Chapter 3: Configuration (32B Config)】
- FP8 base model loading (critical for 24GB VRAM)
- Network settings: Rank 32-64, Alpha 16-32
- Learning rate: Slightly lower than Flux.1 (8e-5)
- VRAM optimization stack: gradient_checkpointing + cache_text_encoder_outputs

【Chapter 4: Training Process】
- Step-by-step execution guide
- Slower iteration speed expected (32B model is massive)
- Loss reference: 0.08-0.12 stable range
- Epoch recommendations: 8 epochs max (32B learns very fast)

【Chapter 5: Pruning & Compatibility】
- File size: ~40MB for Rank 32 (similar to Flux.1)
- CRITICAL: Flux.2 LoRAs NOT compatible with Flux.1
- Clear labeling conventions for model files

【Chapter 6: Testing (Hex Codes & JSON)】
- ComfyUI Flux.2 workflow setup
- Hex-code color steering in prompts
- Resolution recommendations: 1536x1536+ for 4MP benefits
- Weight recommendations: 0.8 starting point

Bonus: FAQ & Troubleshooting
- Immediate OOM solutions (FP8 + caching)
- Slow training expectations (32B reality check)
- Artifact fixes (learning rate sensitivity)

Technical Requirements:
- NVIDIA GPU with 24GB+ VRAM (32GB+ recommended)
- 64-128GB System RAM
- Windows with Kohya_ss 2025 version
- 100GB+ free storage space

Package Contents:
- Flux.2 (FLEX2) LoRA Training Complete Tutorial.html (formatted tutorial)
- Flux.2 (FLEX2) LoRA Training Complete Tutorial.md (markdown source)

Who This Is For:
- Cutting-edge enthusiasts wanting the latest model technology
- Professional studios needing 4MP native resolution
- Users with high-end hardware (24GB+ VRAM, 64GB+ RAM)
- Artists seeking Hex-code color control capabilities