{
  "scope": "Installed version standard options; custom nodes disabled",
  "comfy_commit": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
  "forge_commit": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
  "samplers": [
    {
      "id": "euler",
      "display_name": "Euler",
      "family": "Euler",
      "description": "現在点の予測から一度で次のノイズ量へ進む一次法。基準を作りやすく、1区間あたり通常1回のモデル評価。",
      "aliases": [
        "k_euler"
      ],
      "forge_name": "Euler",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "euler_cfg_pp",
      "display_name": "Euler CFG++",
      "family": "Euler",
      "description": "Eulerの更新に無条件予測を使うCFG++を組み込む。通常CFGの数値をそのまま最適値と見なさず、低CFGでも比較する。",
      "aliases": [
        "euler_cfg_pp"
      ],
      "forge_name": "Euler CFG++",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "euler_ancestral",
      "display_name": "Euler a",
      "family": "Euler",
      "description": "Eulerにancestralのノイズ注入を加える。Seedを固定してもステップ配分を変えると注入の経路が変わる。",
      "aliases": [
        "k_euler_a",
        "k_euler_ancestral"
      ],
      "forge_name": "Euler a",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "euler_ancestral_cfg_pp",
      "display_name": "Euler a CFG++",
      "family": "Euler",
      "description": "ancestral更新とCFG++を組み合わせる。通常のEuler aとの違いは誘導の扱いで、別の画風モデルではない。",
      "aliases": [
        "euler_ancestral_cfg_pp"
      ],
      "forge_name": "Euler a CFG++",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "heun",
      "display_name": "Heun",
      "family": "Heun",
      "description": "Eulerで仮に進んだ先でも傾きを調べ、二つの傾きで補正する二次法。最終区間以外は通常2回評価。",
      "aliases": [
        "k_heun"
      ],
      "forge_name": "Heun",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "heunpp2",
      "display_name": "Heun++ 2",
      "family": "Heun",
      "description": "先の二点まで予測して重み付き補正を行うHeun派生。通常のHeunより評価回数が増え、同じStepsでも重い。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "exp_heun_2_x0",
      "display_name": "Exponential Heun 2 x0",
      "family": "指数積分・その他",
      "description": "データ予測x0と、ノイズ量を対数へ変換した時間座標を使う二次の指数Heun法。SEEDS-2の決定論的な条件を使う実装で、通常Heunとは計算座標が違う。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "exp_heun_2_x0_sde",
      "display_name": "Exponential Heun 2 x0 SDE",
      "family": "指数積分・その他",
      "description": "指数Heunの確率版。実装ではSEEDS-2をr=1で呼び、追加ノイズを含む。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpm_2",
      "display_name": "DPM2",
      "family": "DPM",
      "description": "ノイズ量の幾何平均にある中間点を評価する二次法。二つのモデル評価を使い、DPM++ 2Mとは履歴の使い方が違う。",
      "aliases": [
        "k_dpm_2"
      ],
      "forge_name": "DPM2",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpm_2_ancestral",
      "display_name": "DPM2 a",
      "family": "DPM",
      "description": "DPM2の中間点計算にancestralノイズを加える。通常版と比較するときは計算時間と構図変化の両方を見る。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "lms",
      "display_name": "LMS",
      "family": "多段法",
      "description": "Linear Multistep。過去の傾きを最大4つ保持し、区間ごとの係数で合成する。起動直後は履歴が少なく低い次数から始まる。",
      "aliases": [
        "k_lms"
      ],
      "forge_name": "LMS",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpm_fast",
      "display_name": "DPM fast",
      "family": "DPM",
      "description": "DPM-Solverの1〜3次の更新へ計算予算を振り分ける。渡されたsigma列の内点をそのまま順に使う方式ではない。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpm_adaptive",
      "display_name": "DPM adaptive",
      "family": "DPM",
      "description": "予測誤差と許容値から刻み幅を自動調整するDPM-Solver。表示Stepsが計算回数を直接決めず、20・30・40は同じ結果になることもある。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_2s_ancestral",
      "display_name": "DPM++ 2S a",
      "family": "DPM++",
      "description": "DPM++の二次single-step法にancestralノイズを加える。各区間の中間評価を使うため、履歴を使う2Mと同じ速さとは限らない。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_2s_ancestral_cfg_pp",
      "display_name": "DPM++ 2S a CFG++",
      "family": "DPM++",
      "description": "2S ancestralの誘導をCFG++へ変更した派生。2M CFG++とは区間内評価と履歴利用の違いがある。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_sde",
      "display_name": "DPM++ SDE",
      "family": "DPM++",
      "description": "データ予測で確率微分方程式を解き、区間内の追加評価とBrownianノイズを使う。2M SDEと同一ではない。",
      "aliases": [
        "k_dpmpp_sde"
      ],
      "forge_name": "DPM++ SDE",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_sde_gpu",
      "display_name": "DPM++ SDE GPU",
      "family": "DPM++",
      "description": "DPM++ SDEと同じ更新式でBrownianノイズ生成をGPU側へ移す。CPU版と同じSeedでも同じ画像は保証されない。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_2m",
      "display_name": "DPM++ 2M",
      "family": "DPM++",
      "description": "データ予測を使う二次multistep法。直前の予測を再利用するため、通常は各区間1回の評価で進む。",
      "aliases": [
        "k_dpmpp_2m"
      ],
      "forge_name": "DPM++ 2M",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_2m_cfg_pp",
      "display_name": "DPM++ 2M CFG++",
      "family": "DPM++",
      "description": "2Mの履歴更新にCFG++を導入。通常の2MにKarrasを付ける設定とは役割が違う。",
      "aliases": [
        "dpmpp_2m_cfg_pp"
      ],
      "forge_name": "DPM++ 2M CFG++",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_2m_sde",
      "display_name": "DPM++ 2M SDE",
      "family": "DPM++",
      "description": "2Mの履歴利用にSDEの確率項を加える。標準の補正式はmidpointで、Heun派生と区別する。",
      "aliases": [
        "k_dpmpp_2m_sde"
      ],
      "forge_name": "DPM++ 2M SDE",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_2m_sde_gpu",
      "display_name": "DPM++ 2M SDE GPU",
      "family": "DPM++",
      "description": "2M SDEのBrownianノイズ生成先をGPUへ移す派生。画像と時間の両方を独立して計測する。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_2m_sde_heun",
      "display_name": "DPM++ 2M SDE Heun",
      "family": "DPM++",
      "description": "2M SDE内の履歴補正をmidpointからHeun型へ替える。通常のHeunを丸ごと追加して2倍評価するという意味ではない。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_2m_sde_heun_gpu",
      "display_name": "DPM++ 2M SDE Heun GPU",
      "family": "DPM++",
      "description": "Heun型の2M SDE補正とGPUでのBrownianノイズ生成を組み合わせる。末尾のGPUは品質等級ではない。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_3m_sde",
      "display_name": "DPM++ 3M SDE",
      "family": "DPM++",
      "description": "三次multistepの確率法。過去二つの予測も使うが、各区間で必ず3回評価するわけではない。",
      "aliases": [
        "k_dpmpp_3m_sde"
      ],
      "forge_name": "DPM++ 3M SDE",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "dpmpp_3m_sde_gpu",
      "display_name": "DPM++ 3M SDE GPU",
      "family": "DPM++",
      "description": "3M SDEのノイズ生成をGPUへ移す。履歴がそろう前は低い次数で進む点はCPU版と共通。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "ddpm",
      "display_name": "DDPM",
      "family": "確率法・蒸留",
      "description": "Denoising Diffusion Probabilistic Models由来の逆過程更新。各区間で確率的にノイズを加える。元の学習過程の説明と少ステップ品質は分ける。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "lcm",
      "display_name": "LCM",
      "family": "確率法・蒸留",
      "description": "Latent Consistency Model向けの更新。予測画像へ次のノイズ量を加える。一般モデルの比較出力は診断用で、LCM蒸留モデルの低ステップ性能を表さない。",
      "aliases": [
        "k_lcm"
      ],
      "forge_name": "LCM",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "ipndm",
      "display_name": "iPNDM",
      "family": "多段法",
      "description": "improved Pseudo Numerical Methods。過去のノイズ予測をAdams–Bashforth型に合成する。最大次数4で、履歴を貯めてから高次になる。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "ipndm_v",
      "display_name": "iPNDM-V",
      "family": "多段法",
      "description": "iPNDMに対し可変の刻み幅を係数へ反映する派生。Vをモデルのv-prediction指定と読み替えない。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "deis",
      "display_name": "DEIS",
      "family": "指数積分・その他",
      "description": "Diffusion Exponential Integrator Sampler。ノイズ予測の履歴と積分係数を使う指数積分法で、設定するsigma列も結果に影響する。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "cfgpp_ud10_ab",
      "display_name": "CFG++ UD10 AB",
      "family": "多段法",
      "description": "CFG++の履歴派生。対象コードは履歴重み0.25、現在の無条件項1.0、無条件履歴0.1を使う。UD10の語源を推測せず、実装定数で区別する。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "res_multistep",
      "display_name": "RES Multistep",
      "family": "指数積分・その他",
      "description": "Refined Exponential Solverの多段実装。過去のデータ予測と指数積分の係数を使い、通常版はeta=0。",
      "aliases": [
        "res_multistep"
      ],
      "forge_name": "Res Multistep",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "res_multistep_cfg_pp",
      "display_name": "RES Multistep CFG++",
      "family": "指数積分・その他",
      "description": "RESの多段更新へCFG++を組み込む。履歴利用は共通で、誘導の計算が異なる。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "res_multistep_ancestral",
      "display_name": "RES Multistep a",
      "family": "指数積分・その他",
      "description": "RES多段法にeta=1のancestralノイズを加える。通常版のeta=0と比較し、ステップ増加による構図の差も見る。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "res_multistep_ancestral_cfg_pp",
      "display_name": "RES Multistep a CFG++",
      "family": "指数積分・その他",
      "description": "RESのancestral更新とCFG++の併用。低CFGを含む個別調整が必要で、通常版と同じ最適設定とは限らない。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "gradient_estimation",
      "display_name": "Gradient Estimation",
      "family": "指数積分・その他",
      "description": "現在と直前のデータ予測から勾配を推定し、ge_gamma=2で補正する。実装では各区間1回のモデル評価で履歴を再利用する。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "gradient_estimation_cfg_pp",
      "display_name": "Gradient Estimation CFG++",
      "family": "指数積分・その他",
      "description": "勾配推定法のCFG++版。通常版と同じge_gammaでも無条件予測の扱いが違い、CFGの相性を別に確認する。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "er_sde",
      "display_name": "ER SDE",
      "family": "指数積分・その他",
      "description": "Extended Reverse-Time SDEの解法。最大3段の履歴補正とノイズ量の変換を使う。2M SDEの単なる別名ではない。",
      "aliases": [
        "er_sde"
      ],
      "forge_name": "ER SDE",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "seeds_2",
      "display_name": "SEEDS 2",
      "family": "指数積分・その他",
      "description": "Stochastic Explicit Exponential Derivative-free Solversの2段法。中間点を評価して確率項を加える。数字はSeedの個数ではない。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "seeds_3",
      "display_name": "SEEDS 3",
      "family": "指数積分・その他",
      "description": "SEEDSの3段法。区間内の二つの中間点を利用し、2段法より評価が増える。高次数が必ず高画質を意味するわけではない。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "sa_solver",
      "display_name": "SA Solver",
      "family": "多段法",
      "description": "Stochastic Adams Solver。確率過程の予測・補正に過去の評価を使う。対象版はpredictor_order=3、corrector_order=4。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "sa_solver_pece",
      "display_name": "SA Solver PECE",
      "family": "多段法",
      "description": "Predict–Evaluate–Correct–Evaluateの順で追加評価するSA Solver派生。通常版より計算を使って補正した状態を再評価する。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/k_diffusion/sampling.py"
    },
    {
      "id": "ddim",
      "display_name": "DDIM",
      "family": "確率法・蒸留",
      "description": "Denoising Diffusion Implicit Modelsという名称だが、このComfyUIの選択肢はEulerへ渡す互換経路。Forgeの独立DDIM実装と同一視しない。",
      "aliases": [
        "ddim"
      ],
      "forge_name": "DDIM",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同名だがComfyUIはEuler経路、Forgeは独立DDIM実装",
      "implementation": "comfy/samplers.py"
    },
    {
      "id": "uni_pc",
      "display_name": "UniPC",
      "family": "多段法",
      "description": "Unified Predictor-Corrector。予測と補正を組み合わせる多段法。対象版の基本変種はbh1で、始点付近と終端では次数が変わる。",
      "aliases": [
        "unipc"
      ],
      "forge_name": "UniPC",
      "environment": [
        "ComfyUI",
        "Forge Neo"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/samplers.py"
    },
    {
      "id": "uni_pc_bh2",
      "display_name": "UniPC bh2",
      "family": "多段法",
      "description": "UniPCの補正関数B(h)をbh2へ変更する派生。同じStepsでも係数が違うため、通常UniPCと出力を照合する。",
      "aliases": [],
      "forge_name": null,
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "source_id": "comfyui-sampling-20261018",
      "cross_ui_relation": "同系統の名称対応。既定値・乱数・終端処理の一致は保証しない",
      "implementation": "comfy/samplers.py"
    }
  ],
  "schedulers": [
    {
      "id": "normal",
      "display_name": "Normal",
      "description": "モデルの時間座標を等間隔に取り、その時間からsigmaへ戻す。sigmaの差そのものは等間隔にならず、対応はモデル依存。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [
        "Normal"
      ],
      "forge_id": "normal",
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": true
    },
    {
      "id": "simple",
      "display_name": "Simple",
      "description": "モデルが保持するsigma配列を逆順に一定の添字間隔で拾う。端点や丸めがNormalと異なる。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [
        "Simple"
      ],
      "forge_id": "simple",
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": true
    },
    {
      "id": "sgm_uniform",
      "display_name": "SGM Uniform",
      "description": "時間座標をsteps+1個に分け、最後の時刻を除いてからゼロを付ける。末尾のUniformはsigma自体の等間隔を意味しない。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [
        "SGM Uniform",
        "SGMUniform"
      ],
      "forge_id": "sgm_uniform",
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": true
    },
    {
      "id": "karras",
      "display_name": "Karras",
      "description": "sigmaの1/rho乗を等間隔にしてrho乗で戻す。対象版のrhoは7。sigmaそのものの線形配置でも、必ず高画質になる保証でもない。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [
        "Karras"
      ],
      "forge_id": "karras",
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": true
    },
    {
      "id": "exponential",
      "display_name": "Exponential",
      "description": "log(sigma)を等間隔にする。隣り合うsigmaの比率が一定になり、減る量が一定になるわけではない。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [
        "Exponential"
      ],
      "forge_id": "exponential",
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": true
    },
    {
      "id": "ddim_uniform",
      "display_name": "DDIM Uniform",
      "description": "学習側のsigma配列を整数の添字間隔で取り出す。丸めで実際の区間数が指定Stepsとずれることがある。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [
        "DDIM"
      ],
      "forge_id": "ddim",
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": true
    },
    {
      "id": "beta",
      "display_name": "Beta",
      "description": "Beta分布の逆累積分布で時間の添字を選ぶ。ComfyUI標準はalpha=beta=0.6。丸めによる重複は除く。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [
        "Beta"
      ],
      "forge_id": "beta",
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": true
    },
    {
      "id": "linear_quadratic",
      "display_name": "Linear Quadratic",
      "description": "線形区間と二次曲線の区間をつなぐ。対象版はthreshold_noise=0.025、線形側は全区間の半分。汎用SDXLでの見た目は実画像で判断する。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [
        "Linear Quadratic"
      ],
      "forge_id": "linear_quadratic",
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": true
    },
    {
      "id": "kl_optimal",
      "display_name": "KL Optimal",
      "description": "KLは分布のずれを測るKullback–Leiblerダイバージェンスに由来する名称。atan(sigma)を等間隔にしてtanで戻す。元の理論の前提を離れた全モデルでの最適性は保証しない。ComfyUIとForgeは端点の取り方も異なる。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [
        "KL Optimal"
      ],
      "forge_id": "kl_optimal",
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": true
    },
    {
      "id": "ays_sdxl",
      "display_name": "Align Your Steps（SDXL）",
      "description": "モデル系列別に選んだ既定の計算点を使う。ComfyUI標準の別ノードで、指定Stepsへ対数補間する。Forgeの実装は配列と補間数が異なる。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [
        "Align Your Steps"
      ],
      "forge_id": "align_your_steps",
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": false
    },
    {
      "id": "polyexponential",
      "display_name": "Polyexponential",
      "description": "log(sigma)側の間隔へrho乗をかける。rho=1ならExponentialと同じ式になる。標準の別ノード。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [
        "Polyexponential"
      ],
      "forge_id": "polyexponential",
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": false
    },
    {
      "id": "laplace",
      "display_name": "Laplace",
      "description": "Laplace分布の形でlog(sigma)を配置する。muは中心、betaは広がり。端点範囲へ制限するため重複点にも注意する。標準の別ノード。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [],
      "forge_id": null,
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": false
    },
    {
      "id": "vp",
      "display_name": "VP",
      "description": "Variance Preservingの連続過程に基づきbeta_d・beta_min・eps_sからsigma列を作る。端点が通常SDXLの既定範囲と異なる別条件。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [],
      "forge_id": null,
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": false
    },
    {
      "id": "gits",
      "display_name": "GITS",
      "description": "既定の計算点テーブルを係数とStepsで選び、長い列へは補間する標準ノード。coeff=1.2を使用。モデル横断の最適保証ではない。",
      "family": "スケジューラー",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "aliases": [],
      "forge_id": null,
      "source_id": "comfyui-sampling-20261018",
      "ksampler_option": false
    }
  ],
  "forge_samplers": [
    {
      "display_name": "DPM++ 2M",
      "function": "sample_dpmpp_2m",
      "aliases": [
        "k_dpmpp_2m"
      ],
      "options": {
        "scheduler": "karras"
      },
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_dpmpp_2m",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "DPM++ SDE",
      "function": "sample_dpmpp_sde",
      "aliases": [
        "k_dpmpp_sde"
      ],
      "options": {
        "scheduler": "karras",
        "second_order": true,
        "brownian_noise": true
      },
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_dpmpp_sde",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "DPM++ 2M SDE",
      "function": "sample_dpmpp_2m_sde",
      "aliases": [
        "k_dpmpp_2m_sde"
      ],
      "options": {
        "scheduler": "exponential",
        "brownian_noise": true
      },
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_dpmpp_2m_sde",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "DPM++ 3M SDE",
      "function": "sample_dpmpp_3m_sde",
      "aliases": [
        "k_dpmpp_3m_sde"
      ],
      "options": {
        "scheduler": "exponential",
        "discard_next_to_last_sigma": true,
        "brownian_noise": true
      },
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_dpmpp_3m_sde",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "DPM++ 2s a RF",
      "function": "sample_dpmpp_2s_ancestral_RF",
      "aliases": [
        "sample_dpmpp_2s_ancestral_RF"
      ],
      "options": {},
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_dpmpp_2s_ancestral_RF",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "Euler a",
      "function": "sample_euler_ancestral",
      "aliases": [
        "k_euler_a",
        "k_euler_ancestral"
      ],
      "options": {
        "uses_ensd": true
      },
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_euler_ancestral",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "Euler",
      "function": "sample_euler",
      "aliases": [
        "k_euler"
      ],
      "options": {},
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_euler",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "ER SDE",
      "function": "sample_er_sde",
      "aliases": [
        "er_sde"
      ],
      "options": {},
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_er_sde",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "LCM",
      "function": "sample_lcm",
      "aliases": [
        "k_lcm"
      ],
      "options": {},
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_lcm",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "LMS",
      "function": "sample_lms",
      "aliases": [
        "k_lms"
      ],
      "options": {},
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_lms",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "Heun",
      "function": "sample_heun",
      "aliases": [
        "k_heun"
      ],
      "options": {
        "second_order": true
      },
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_heun",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "DPM2",
      "function": "sample_dpm_2",
      "aliases": [
        "k_dpm_2"
      ],
      "options": {
        "scheduler": "karras",
        "discard_next_to_last_sigma": true,
        "second_order": true
      },
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_dpm_2",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "Res Multistep",
      "function": "sample_res_multistep",
      "aliases": [
        "res_multistep"
      ],
      "options": {},
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_res_multistep",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "Kohaku LoNyu Yog",
      "function": "sample_Kohaku_LoNyu_Yog",
      "aliases": [
        "Kohaku_LoNyu_Yog"
      ],
      "options": {},
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sample_Kohaku_LoNyu_Yog",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "Restart",
      "function": "sd_samplers_extra.restart_sampler",
      "aliases": [
        "restart"
      ],
      "options": {
        "scheduler": "karras",
        "second_order": true
      },
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sd_samplers_extra.restart_sampler",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "UniPC",
      "function": "sd_samplers_extra.sample_unipc",
      "aliases": [
        "unipc"
      ],
      "options": {
        "discard_next_to_last_sigma": true
      },
      "source_file": "modules/sd_samplers_kdiffusion.py",
      "id": "sd_samplers_extra.sample_unipc",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "DDIM",
      "function": "sd_samplers_timesteps_impl.ddim",
      "aliases": [
        "ddim"
      ],
      "options": {},
      "source_file": "modules/sd_samplers_timesteps.py",
      "id": "sd_samplers_timesteps_impl.ddim",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "PLMS",
      "function": "sd_samplers_timesteps_impl.plms",
      "aliases": [
        "plms"
      ],
      "options": {},
      "source_file": "modules/sd_samplers_timesteps.py",
      "id": "sd_samplers_timesteps_impl.plms",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "DPM++ 2M CFG++",
      "function": "sample_dpmpp_2m_cfg_pp",
      "aliases": [
        "dpmpp_2m_cfg_pp"
      ],
      "options": {
        "scheduler": "karras"
      },
      "source_file": "modules_forge/forge_alter_samplers.py",
      "id": "sample_dpmpp_2m_cfg_pp",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "Euler a CFG++",
      "function": "sample_euler_ancestral_cfg_pp",
      "aliases": [
        "euler_ancestral_cfg_pp"
      ],
      "options": {
        "uses_ensd": true
      },
      "source_file": "modules_forge/forge_alter_samplers.py",
      "id": "sample_euler_ancestral_cfg_pp",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    },
    {
      "display_name": "Euler CFG++",
      "function": "sample_euler_cfg_pp",
      "aliases": [
        "euler_cfg_pp"
      ],
      "options": {},
      "source_file": "modules_forge/forge_alter_samplers.py",
      "id": "sample_euler_cfg_pp",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "sampler",
      "source_id": "forge-neo-samplers-20261018"
    }
  ],
  "forge_schedulers": [
    {
      "id": "automatic",
      "display_name": "Automatic",
      "function": "None",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "karras",
      "display_name": "Karras",
      "function": "k_diffusion.sampling.get_sigmas_karras",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "exponential",
      "display_name": "Exponential",
      "function": "k_diffusion.sampling.get_sigmas_exponential",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "polyexponential",
      "display_name": "Polyexponential",
      "function": "k_diffusion.sampling.get_sigmas_polyexponential",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "normal",
      "display_name": "Normal",
      "function": "normal_scheduler",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "simple",
      "display_name": "Simple",
      "function": "simple_scheduler",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "uniform",
      "display_name": "Uniform",
      "function": "uniform",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "sgm_uniform",
      "display_name": "SGM Uniform",
      "function": "sgm_uniform",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": [
        "SGMUniform"
      ]
    },
    {
      "id": "linear_quadratic",
      "display_name": "Linear Quadratic",
      "function": "linear_quadratic",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "kl_optimal",
      "display_name": "KL Optimal",
      "function": "kl_optimal",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "ddim",
      "display_name": "DDIM",
      "function": "ddim_scheduler",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "align_your_steps",
      "display_name": "Align Your Steps",
      "function": "get_align_your_steps_sigmas",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "beta",
      "display_name": "Beta",
      "function": "beta_scheduler",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "turbo",
      "display_name": "Turbo",
      "function": "turbo_scheduler",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "bong_tangent",
      "display_name": "Bong Tangent",
      "function": "bong_tangent_scheduler",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "flow_match",
      "display_name": "FlowMatchEulerDiscrete",
      "function": "flow_match_euler_discrete_scheduler",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    },
    {
      "id": "flux2",
      "display_name": "Flux2",
      "function": "flux2_scheduler",
      "environment": [
        "Forge Neo"
      ],
      "version": "f7f231496b2dd46703f8da87e2e83b45a54ba73b",
      "classification": "scheduler_option",
      "source_id": "forge-neo-schedulers-20261018",
      "aliases": []
    }
  ],
  "model_specific_nodes": [
    {
      "id": "SDTurboScheduler",
      "description": "SD-Turboの少ステップ時刻列。対応する蒸留モデルが未導入のため適合実験は未実施。通常モデルの20〜40Steps比較へ混ぜない。",
      "display_name": "SDTurboScheduler",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "classification": "model-specific scheduler",
      "source_id": "comfyui-model-schedules-20261018",
      "test_status": "no_matching_model"
    },
    {
      "id": "Ideogram4Scheduler",
      "description": "Ideogram 4向け。モデル固有の時刻変換と解像度条件を使うためSDXL比較の対象外。",
      "display_name": "Ideogram4Scheduler",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "classification": "model-specific scheduler",
      "source_id": "comfyui-model-schedules-20261018",
      "test_status": "no_matching_model"
    },
    {
      "id": "Flux2Scheduler",
      "description": "Flux.2向けに画像の系列長とStepsから時間をシフトする。SDXLのsigma列の代替にはしない。",
      "display_name": "Flux2Scheduler",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "classification": "model-specific scheduler",
      "source_id": "comfyui-model-schedules-20261018",
      "test_status": "no_matching_model"
    },
    {
      "id": "LTXVScheduler",
      "description": "LTX-Video向けのシフト付き時刻列。トークン数や終端条件を使う動画モデル用。",
      "display_name": "LTXVScheduler",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "classification": "model-specific scheduler",
      "source_id": "comfyui-model-schedules-20261018",
      "test_status": "no_matching_model"
    },
    {
      "id": "OptimalStepsScheduler",
      "description": "FLUX・Wan・Chroma別の既定点を対数補間する。名称がOptimalでもSDXLの適合性を示さない。",
      "display_name": "OptimalStepsScheduler",
      "environment": [
        "ComfyUI"
      ],
      "version": "98cc0bf84b7cb1383654c5b402e7612d875d6d22",
      "classification": "model-specific scheduler",
      "source_id": "comfyui-model-schedules-20261018",
      "test_status": "no_matching_model"
    }
  ]
}