Boogu Edit on Qwen Edit Demos 2

Continue evaluating Boogu Edit model against the test scenarios from the Qwen Edit demo. By applying Boogu Edit to nearly identical input images and editing prompts, we assess its relative output quality, instruction following ability and overall image output editing capabilities compared to Qwen Edit output. Refer to https://huggingface.co/Qwen/Qwen-Image-Edit-2511

Files

boogu_image_edit_turbo_hotfix_1k_20260708_int8_convrot.safetensors

SHA256: b0bdd36eb559d62e8169e16a1e8b4438f8be17fc092ca9d43a2d8460c1d79e6a

Lora Integration - Relighting

• 1. Test 7
• 2. Test 8

Lora Integration - Novel View Synthesis

• 3. Test 9
• 4. Test 10

Industry Design

• 5. Test 11
• 5. Test 12
• 5. Test 13
• 5. Test 14

Test 7

Input image:

Prompt text:

柔光,使用柔和光线对图片进行重新照明

Output image:

---

Test 8

Input image:

Prompt text:

柔光,使用柔和光线对图片进行重新照明

Boogu Prompt text:

对图片进行重新照明

Output image:

---

Test 9

Input image:

Prompt text:

将镜头平移至桌面特写

Boogu Prompt text:

把镜头移到桌面上的树枝造型四头烛台进行放大特写

Output image:

---

Test 10

Input image:

Prompt text:

将镜头向左旋转30度

Boogu Prompt text:

暂时没办法。

Output image:

---

Test 11

Input image:

Prompt text:

把图中的汽车变成灰色

Output image:

---

Test 12

Input image:

Prompt text:

把图片中5kg砝码的拉提手部分去掉

Output image:

---