--- license: creativeml-openrail-m base_model: Lykon/dreamshaper-7 base_model_relation: quantized pipeline_tag: text-to-image library_name: gguf language: - en tags: - text-to-image - image-to-image - stable-diffusion - ip-adapter - dreamshaper - reference-image - identity - gguf - stable-diffusion-cpp - tokforge - on-device - image-generation - sd1.5 --- # SD1.5 IP-Adapter Plus Face (GGUF, safetensors) DreamShaper 7 with the IP-Adapter plus-face adapter and its image encoder, so TokForge can put the person from a reference photo into any scene you describe. Published by [TokForge](https://tokforge.ai) for on-device inference. | | | |---|---| | **What it is** | Image generator (reference photo) | | **Format** | f16 GGUF base for stable-diffusion.cpp, plus IP-Adapter and image encoder safetensors | | **Runs on** | Android phones (stable-diffusion.cpp, CPU or Adreno OpenCL GPU) | | **Phone class** | 8 GB RAM or more (TokForge catalog) | | **Download** | 4.84 GB (5 files) | | **Base model** | [Lykon/dreamshaper-7](https://huggingface.co/Lykon/dreamshaper-7), [h94/IP-Adapter](https://huggingface.co/h94/IP-Adapter) | | **License** | Several: [CreativeML OpenRAIL-M](https://huggingface.co/spaces/CompVis/stable-diffusion-license) plus Lykon's Civitai terms (base), Apache-2.0 (adapter), MIT (image encoder) | ## TokForge - **Website:** https://tokforge.ai - **Discord:** https://discord.gg/Acv3CBtfVm - **Google Play:** https://play.google.com/store/apps/details?id=dev.tokforge - **iOS TestFlight:** https://testflight.apple.com/join/jnufjzRr Runs on-device in the TokForge app. ## Use it in TokForge 1. Install TokForge from [Google Play](https://play.google.com/store/apps/details?id=dev.tokforge). 2. On Android, open **Models > Recommended > Image** and download **SD1.5 IP-Adapter (Reference Identity)**. TokForge lists it on phones with 8 GB RAM or more. 3. In a chat, attach a clear photo of the person's face as a reference, then describe the scene (for example *"me as a superhero flying over New York"*). The adapter keeps the face; the prompt drives everything else. 4. The model runs on the phone; once it is downloaded, making pictures needs no connection. Collections: [TokForge GGUF Models](https://huggingface.co/collections/darkmaniac7/tokforge-gguf-models-6af3fdd830c420213916bda1), [TokForge Image Models (Android)](https://huggingface.co/collections/darkmaniac7/tokforge-image-models-android-6a2195edbecaee6d762b71b4). ## Files | File | Size | What it is | |---|---:|---| | `ip-adapter-plus-face_sd15.safetensors` | 98.2 MB | IP-Adapter plus-face for SD1.5 (h94/IP-Adapter, Apache-2.0) | | `ip_adapter_clip_vision_vith.safetensors` | 2.53 GB | OpenCLIP ViT-H-14 image encoder for the reference photo (MIT) | | `manifest.json` | 1.29 KB | Render defaults, licences, sizes and md5 of the three model files | | `sd15-base-f16.gguf` | 2.21 GB | DreamShaper 7 base: CLIP text encoder, UNet and VAE in one f16 GGUF | | `MD5SUMS` | 199 B | md5 of the three model files | `manifest.json` and `MD5SUMS` carry the integrity hashes and render defaults. ## About the model The **reference-identity** image route for the TokForge Android app. Attach a photo of a person, then render **that person** in any scene. The **plus-face** IP-Adapter transfers the **face only** while the **prompt drives the whole scene**. This bundle runs on the on-device [`stable-diffusion.cpp`](https://github.com/leejet/stable-diffusion.cpp) GGUF engine (TokForge's IP-Adapter port) on **CPU** and **Adreno OpenCL**. SD1.5 is light enough for any 8 GB+ phone, the broadest-reach identity tier (lighter than the SDXL PhotoMaker and SDXL IP-Adapter tiers). | File | License | Contents | |------|---------|----------| | `sd15-base-f16.gguf` | CreativeML-OpenRAIL-M | **DreamShaper-7** (SD1.5 realistic finetune), CLIP text encoder + UNet + VAE in one **f16** GGUF | | `ip-adapter-plus-face_sd15.safetensors` | Apache-2.0 | IP-Adapter **plus-face** (`h94/IP-Adapter`), 16-token Resampler + decoupled cross-attn | | `ip_adapter_clip_vision_vith.safetensors` | MIT | OpenCLIP **ViT-H-14** image encoder (the plus-face path needs ViT-H, not bigG) | ### Why this base, and why f16 The base is the **standard, non-LCM DreamShaper-7**, the same realistic SD1.5 finetune TokForge ships on its other image tiers. It is converted at **f16** (full precision) so the IP-Adapter's decoupled cross-attention and the face Resampler keep **subject quality** high; a `q4_0` base weakens the transferred identity. ### Why plus-face The **plus-face** Resampler extracts the **face only** (16 tokens from the ViT-H penultimate hidden state), so identity is preserved while the **prompt** controls the scene. The base `ip-adapter_sd15` projects the whole pooled CLIP embedding (4 tokens) and carries the reference's whole scene with it. The TokForge sd.cpp IP-Adapter loader auto-detects plus-face by the presence of `image_proj.latents`. ## Render settings | Setting | Value | |---------|-------| | sampler | `euler_a` | | scheduler | `discrete` | | steps | `30` (full quality; fewer = faster) | | cfg-scale | `7.0` | | ip-adapter-scale | `0.6` (about 0.5 to 0.6 keeps the scene with recognizable identity; about 0.8 reconstructs the reference) | | resolution | `512x512` (SD1.5 native) | TokForge runs it through its IP-Adapter port of stable-diffusion.cpp, equivalent to: ```bash sd -M img_gen \ -m sd15-base-f16.gguf \ -p "as a superhero flying over New York" \ -n "" \ --clip_vision ip_adapter_clip_vision_vith.safetensors \ --ip-adapter ip-adapter-plus-face_sd15.safetensors \ --ip-adapter-image \ --ip-adapter-scale 0.6 \ --cfg-scale 7.0 --sampling-method euler_a --scheduler discrete \ --steps 30 -H 512 -W 512 ``` ## How it was made - Base converted from `Lykon/dreamshaper-7` (diffusers) to a single f16 GGUF via the TokForge `stable-diffusion.cpp` convert path (`-M convert --type f16`). - Adapter + image encoder copied verbatim from `h94/IP-Adapter` (`models/ip-adapter-plus-face_sd15.safetensors`, `models/image_encoder/model.safetensors`); both files are byte-identical to the upstream ones. ## Limitations - The `--ip-adapter` options belong to TokForge's IP-Adapter port; stock stable-diffusion.cpp builds may not have them. - Plus-face carries the face only, so keep the face visible and unobstructed for a recognizable result. ## License This is an aggregate of three independently-licensed components, and each retains its own license: - **DreamShaper-7 base** (`sd15-base-f16.gguf`): **CreativeML-OpenRAIL-M** ([Lykon/dreamshaper-7](https://huggingface.co/Lykon/dreamshaper-7)). Use must comply with the OpenRAIL-M use-based restrictions. - **IP-Adapter plus-face** (`ip-adapter-plus-face_sd15.safetensors`): **Apache-2.0** ([h94/IP-Adapter](https://huggingface.co/h94/IP-Adapter)). - **OpenCLIP ViT-H-14 image encoder** (`ip_adapter_clip_vision_vith.safetensors`): **MIT** (OpenCLIP / LAION ViT-H-14). > The non-commercial **IP-Adapter-FaceID** / InsightFace path is **NOT** used here, only the > Apache-2.0 base + plus-face adapters from `h94/IP-Adapter`. DreamShaper 7 is also under Lykon's Civitai terms: no selling the model or merges of it. ## Credits - DreamShaper 7 by Lykon ([Civitai](https://civitai.com/models/4384), [Hugging Face](https://huggingface.co/Lykon/dreamshaper-7)). Used under the CreativeML OpenRAIL-M license and the creator's Civitai terms. Lykon's Civitai terms: no selling the model or merges of it. - Stable Diffusion v1.5 by Runway and CompVis ([Hugging Face](https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5)). Used under the CreativeML OpenRAIL-M license. - IP-Adapter by Tencent AI Lab (h94) ([Hugging Face](https://huggingface.co/h94/IP-Adapter)). Used under the Apache-2.0 license. - OpenCLIP ViT-H-14 image encoder by LAION ([Hugging Face](https://huggingface.co/laion/CLIP-ViT-H-14-laion2B-s32B-b79K)). Used under the MIT license. - Converted to GGUF with [stable-diffusion.cpp](https://github.com/leejet/stable-diffusion.cpp) by leejet (MIT).