Clarify X2HDR LoRA requirement
This commit is contained in:
@@ -45,6 +45,8 @@ X2HDR model or LoRA
|
||||
-> Save Image preview PNG, or X2HDR Save EXR for graded HDR
|
||||
```
|
||||
|
||||
Important: these nodes do not convert a normal SDR model output into HDR. `X2HDR PU21 Decode` only performs the inverse transform for an image that is already in the X2HDR/PU21 representation. If the upstream model or LoRA was not trained to generate PU21-encoded HDR images, decoding it with this node will produce incorrect colors and luminance, not real HDR.
|
||||
|
||||
## Installation
|
||||
|
||||
Place this folder under:
|
||||
@@ -69,11 +71,12 @@ ComfyUI already provides the base runtime used by these nodes, including `torch`
|
||||
|
||||
## Typical Workflow
|
||||
|
||||
1. Connect your model's `VAE Decode` `IMAGE` output to `X2HDR PU21 Decode`.
|
||||
2. Send `hdr_image` to `X2HDR Dynamic Range QA` to verify that the result contains useful HDR range.
|
||||
3. Save the decoded linear image with `X2HDR Save EXR`.
|
||||
4. Use `X2HDR Color Grade` for display rendering or creative adjustment.
|
||||
5. Save `graded_display` with normal `Save Image`, or save `graded_linear` with `X2HDR Save EXR` if you need graded HDR output.
|
||||
1. Use an X2HDR/PU21-trained model or LoRA, such as the Krea2 X2HDR LoRA below.
|
||||
2. Connect that workflow's `VAE Decode` `IMAGE` output to `X2HDR PU21 Decode`.
|
||||
3. Send `hdr_image` to `X2HDR Dynamic Range QA` to verify that the result contains useful HDR range.
|
||||
4. Save the decoded linear image with `X2HDR Save EXR`.
|
||||
5. Use `X2HDR Color Grade` for display rendering or creative adjustment.
|
||||
6. Save `graded_display` with normal `Save Image`, or save `graded_linear` with `X2HDR Save EXR` if you need graded HDR output.
|
||||
|
||||
An example workflow scaffold is included at:
|
||||
|
||||
@@ -89,7 +92,7 @@ A Krea2 X2HDR LoRA is available at:
|
||||
https://huggingface.co/F16/x2hdr-krea2
|
||||
```
|
||||
|
||||
Use it with Krea-2-Raw or Krea-2-Turbo, then decode the `VAE Decode` output with `X2HDR PU21 Decode` before saving EXR.
|
||||
Use it with Krea-2-Raw or Krea-2-Turbo, then decode the `VAE Decode` output with `X2HDR PU21 Decode` before saving EXR. The LoRA is required because it teaches the denoiser to output a PU21-encoded HDR representation; the node alone cannot infer HDR data from a normal LDR model output.
|
||||
|
||||
## Decode Defaults
|
||||
|
||||
|
||||
+9
-6
@@ -45,6 +45,8 @@ X2HDR model or LoRA
|
||||
-> Save Image preview PNG,或用 X2HDR Save EXR 保存调色后的 HDR
|
||||
```
|
||||
|
||||
重要:这些节点不能把普通 SDR 模型输出“转换成 HDR”。`X2HDR PU21 Decode` 只负责对已经处在 X2HDR/PU21 表示中的图像做反变换。如果上游模型或 LoRA 没有按 PU21 编码 HDR 的方式训练,直接接这个节点会得到错误的颜色和亮度,不会得到真正的 HDR。
|
||||
|
||||
## 安装
|
||||
|
||||
将本目录放到:
|
||||
@@ -69,11 +71,12 @@ ComfyUI 已经提供这些节点使用的基础运行环境,包括 `torch`、`
|
||||
|
||||
## 典型工作流
|
||||
|
||||
1. 将模型的 `VAE Decode` `IMAGE` 输出接到 `X2HDR PU21 Decode`。
|
||||
2. 将 `hdr_image` 接到 `X2HDR Dynamic Range QA`,确认结果确实包含可用 HDR 范围。
|
||||
3. 用 `X2HDR Save EXR` 保存解码后的线性图像。
|
||||
4. 用 `X2HDR Color Grade` 做显示渲染或创意调色。
|
||||
5. 用普通 `Save Image` 保存 `graded_display`,或用 `X2HDR Save EXR` 保存 `graded_linear` 作为调色后的 HDR 输出。
|
||||
1. 使用按 X2HDR/PU21 方式训练的模型或 LoRA,例如下面的 Krea2 X2HDR LoRA。
|
||||
2. 将该工作流的 `VAE Decode` `IMAGE` 输出接到 `X2HDR PU21 Decode`。
|
||||
3. 将 `hdr_image` 接到 `X2HDR Dynamic Range QA`,确认结果确实包含可用 HDR 范围。
|
||||
4. 用 `X2HDR Save EXR` 保存解码后的线性图像。
|
||||
5. 用 `X2HDR Color Grade` 做显示渲染或创意调色。
|
||||
6. 用普通 `Save Image` 保存 `graded_display`,或用 `X2HDR Save EXR` 保存 `graded_linear` 作为调色后的 HDR 输出。
|
||||
|
||||
示例 workflow scaffold 位于:
|
||||
|
||||
@@ -89,7 +92,7 @@ Krea2 X2HDR LoRA 位于:
|
||||
https://huggingface.co/F16/x2hdr-krea2
|
||||
```
|
||||
|
||||
可配合 Krea-2-Raw 或 Krea-2-Turbo 使用。生成后需要先将 `VAE Decode` 输出接入 `X2HDR PU21 Decode`,再保存 EXR。
|
||||
可配合 Krea-2-Raw 或 Krea-2-Turbo 使用。生成后需要先将 `VAE Decode` 输出接入 `X2HDR PU21 Decode`,再保存 EXR。必须配合该类 LoRA 使用的原因是:LoRA 让 denoiser 学会输出 PU21 编码的 HDR 表示;节点本身无法从普通 LDR 模型输出中推断出 HDR 数据。
|
||||
|
||||
## 默认解码参数
|
||||
|
||||
|
||||
Reference in New Issue
Block a user