CSU-JPG/VPBench

FlowInOne: Unifying Multimodal Generation as Image-in, Image-out Flow Matching

TL;DR: The first vision-centric image-in, image-out image generation model.

🌐 Homepage | 💻 Code | 📄 Paper | 📁 Dataset | 🌏 Benchmark | 🤗 Model

VP-Bench

VP-Bench is the official evaluation benchmark for FlowInOne. It is a rigorously curated benchmark assessing instruction faithfulness, spatial precision, visual realism, and content consistency across eight distinct visual prompting tasks.

Evaluation

Our evaluation scripts are now available on GitHub!

Dataset Subsets

The dataset contains 8 subsets, each corresponding to a distinct visual instruction task:

Subset	Abbrev.	Description
class2image	C2I	Class label rendered in input image → generate corresponding image
text2image	T2I	Text instruction rendered in input image → generate image
text_in_image	TIE	Edit text content within an image
force	FU	Physics-aware force understanding (3 categories)
text_box_control	TBE	Text and bounding box editing
trajectory	TU	Trajectory understanding and prediction
vismarker	VME	Visual marker guided editing (8 categories)
doodles	DE	Doodle-based editing

Dataset Features

• input_image (image): The input visual prompt image (with rendered instruction).
• output_image (image): The ground-truth output image.
• recognized_text (string): The text instruction rendered in the input image (extracted via OCR annotation).
• subset (string): The subset name.
• category (string): Sub-category within a subset (empty string if not applicable).
• image_name (string): The image filename.
• input_relpath (string): Relative path of the input image within the subset.
• output_relpath (string): Relative path of the output image within the subset.
• pair_id (string): Stable SHA1 identifier for each input-output pair.

Loading the Dataset

# class2image
from datasets import load_dataset
ds = load_dataset("CSU-JPG/VPBench", "class2image", split="train")
# text2image
from datasets import load_dataset
ds = load_dataset("CSU-JPG/VPBench", "text2image", split="train")
# text_in_image
from datasets import load_dataset
ds = load_dataset("CSU-JPG/VPBench", "text_in_image", split="train")
# force
from datasets import load_dataset
ds = load_dataset("CSU-JPG/VPBench", "force", split="train")
# text_box_control
from datasets import load_dataset
ds = load_dataset("CSU-JPG/VPBench", "text_box_control", split="train")
# trajectory
from datasets import load_dataset
ds = load_dataset("CSU-JPG/VPBench", "trajectory", split="train")
# vismarker
from datasets import load_dataset
ds = load_dataset("CSU-JPG/VPBench", "vismarker", split="train")
# doodles
from datasets import load_dataset
ds = load_dataset("CSU-JPG/VPBench", "doodles", split="train")
# Load All Subsets
from datasets import load_dataset, concatenate_datasets
subsets = ["class2image", "text2image", "text_in_image", "force",
           "text_box_control", "trajectory", "vismarker", "doodles"]
ds_all = concatenate_datasets([
    load_dataset("CSU-JPG/VPBench", name=s, split="train") for s in subsets
])

Evaluation Results

We evaluate multiple methods on VP-Bench using three state-of-the-art VLM evaluators (Gemini3, GPT-5.2, Qwen3.5) and human judges. The metric is success ratio (higher is better). Total denotes the average success rate across all eight task categories.

Abbreviations: C2I: class-to-image · T2I: text-to-image · TIE: text-in-image edit · FU: force understanding · TBE: text & bbox edit · TU: trajectory understanding · VME: visual marker edit · DE: doodles edit

Evaluator: Gemini3

Method	C2I	T2I	TIE	FU	TBE	TU	VME	DE	Total
Nano Banana (Google, 2025)	.650	.980	.423	.520	.614	.020	.548	.721	.560
Omnigen2 (Wu et al., 2025)	.020	.020	.017	.020	.000	.000	.000	.000	.007
Kontext (Labs et al., 2025)	.050	.020	.048	.007	.000	.020	.010	.000	.019
Qwen-IE-2509 (Wu et al., 2025)	.230	.040	.069	.000	.000	.020	.023	.000	.048
FlowInOne (Ours)	.890	.700	.355	.727	.302	.520	.292	.535	.540

Evaluator: GPT-5.2

Method	C2I	T2I	TIE	FU	TBE	TU	VME	DE	Total
Nano Banana (Google, 2025)	.680	.959	.152	.127	.023	.040	.136	.302	.302
Omnigen2 (Wu et al., 2025)	.110	.020	.000	.000	.000	.000	.000	.023	.019
Kontext (Labs et al., 2025)	.090	.020	.028	.020	.000	.080	.003	.093	.042
Qwen-IE-2509 (Wu et al., 2025)	.240	.120	.080	.020	.022	.060	.020	.047	.076
FlowInOne (Ours)	.850	.800	.079	.500	.116	.240	.083	.465	.392

Evaluator: Qwen3.5

Method	C2I	T2I	TIE	FU	TBE	TU	VME	DE	Total
Nano Banana (Google, 2025)	.600	.959	.386	.367	.257	.040	.321	.744	.469
Omnigen2 (Wu et al., 2025)	.030	.020	.017	.034	.000	.000	.003	.047	.019
Kontext (Labs et al., 2025)	.050	.020	.042	.133	.000	.060	.047	.093	.056
Qwen-IE-2509 (Wu et al., 2025)	.270	.060	.080	.087	.047	.040	.033	.047	.083
FlowInOne (Ours)	.859	.720	.354	.713	.272	.320	.306	.481	.503

Evaluator: Human

Method	C2I	T2I	TIE	FU	TBE	TU	VME	DE	Total
Nano Banana (Google, 2025)	.602	.904	.271	.250	.200	.050	.229	.742	.406
Omnigen2 (Wu et al., 2025)	.000	.000	.000	.000	.000	.000	.000	.000	.000
Kontext (Labs et al., 2025)	.000	.000	.043	.000	.000	.000	.000	.100	.018
Qwen-IE-2509 (Wu et al., 2025)	.067	.000	.029	.000	.000	.000	.000	.000	.012
FlowInOne (Ours)	.800	.645	.242	.705	.255	.280	.255	.400	.449

Citation

If you found our work useful, please consider citing:

@article{

}