# Text-to-Image Generation on ImageNet (Density, Coverage, VS, ClipScore, SPP)

Source: [Text-to-Image Generation on ImageNet (Density, Coverage, VS, ClipScore, SPP)](<https://www.sota2.com/research/sota/text-to-image-generation-on-imagenet-density-coverage-vs-clipscore-spp>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:40:43.335Z

Evaluation of Text-to-Image generation models using the standard ImageNet dataset, assessed using a specific suite of metrics including Density, Coverage, VS, ClipScore, and SPP.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Text-to-Image Generation |
| Dataset (including any stated split) | ImageNet |
| Metric | Density |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 12 of 12 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Density](<https://www.sota2.com/research/sota/text-to-image-generation-on-imagenet-density-coverage-vs-clipscore-spp.md?metric=Density>): Higher is better; unit: SCALAR.

- [Coverage](<https://www.sota2.com/research/sota/text-to-image-generation-on-imagenet-density-coverage-vs-clipscore-spp.md?metric=Coverage>): Higher is better; unit: SCALAR.

- [VS](<https://www.sota2.com/research/sota/text-to-image-generation-on-imagenet-density-coverage-vs-clipscore-spp.md?metric=VS>): Higher is better; unit: SCALAR.

- [ClipScore](<https://www.sota2.com/research/sota/text-to-image-generation-on-imagenet-density-coverage-vs-clipscore-spp.md?metric=ClipScore>): Higher is better; unit: SCALAR.

- [SPP](<https://www.sota2.com/research/sota/text-to-image-generation-on-imagenet-density-coverage-vs-clipscore-spp.md?metric=SPP>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Density | Result date | Report paper |
| --- | --- | --- | --- |
| IG-NeurIPS’24 | 1.389 | Not recorded | 6997ea7209c33900fcd76506 |
| CADS-ICLR’24 | 1.374 | Not recorded | 6997ea7209c33900fcd76506 |
| GASS | 1.164 | Not recorded | 6997ea7209c33900fcd76506 |
| Vanilla Guidance (CFG) | 1.13 | Not recorded | 6997ea7209c33900fcd76506 |
| Vanilla Guidance (CFG) | 1.105 | Not recorded | 6997ea7209c33900fcd76506 |
| SPELL-ICML’25 | 1.105 | Not recorded | 6997ea7209c33900fcd76506 |
| PG-ICLR’24 | 1.103 | Not recorded | 6997ea7209c33900fcd76506 |
| PG-ICLR’24 | 1.039 | Not recorded | 6997ea7209c33900fcd76506 |
| GASS | 1.015 | Not recorded | 6997ea7209c33900fcd76506 |
| SPELL-ICML’25 | 0.912 | Not recorded | 6997ea7209c33900fcd76506 |
| IG-NeurIPS’24 | 0.901 | Not recorded | 6997ea7209c33900fcd76506 |
| CADS-ICLR’24 | 0.897 | Not recorded | 6997ea7209c33900fcd76506 |

### IG-NeurIPS’24

Density: 1.389. Result ID: 699b08845698665ec49b1bc6.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa758c1e209131cea6889d.md>)

Recorded evaluation configuration: {"Base Model":"SD3-M"}. This record does not assert independent reproduction.

### CADS-ICLR’24

Density: 1.374. Result ID: 699b08845698665ec49b1bc5.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73261e209131cea5ab3e.md>)

[Implementation](<https://github.com/CompVis/latent-diffusion>)

Recorded evaluation configuration: {"Base Model":"SD3-M"}. This record does not assert independent reproduction.

### GASS

Density: 1.164. Result ID: 699b08845698665ec49b1bc8.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

Recorded evaluation configuration: {"Base Model":"SD3-M"}. This record does not assert independent reproduction.

### Vanilla Guidance (CFG)

Density: 1.13. Result ID: 699b08845698665ec49b1bbd.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d1e1e209131cea37e43.md>)

Recorded evaluation configuration: {"Base Model":"SD2.1"}. This record does not assert independent reproduction.

### Vanilla Guidance (CFG)

Density: 1.105. Result ID: 699b08845698665ec49b1bc3.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6d1e1e209131cea37e43.md>)

Recorded evaluation configuration: {"Base Model":"SD3-M"}. This record does not assert independent reproduction.

### SPELL-ICML’25

Density: 1.105. Result ID: 699b08845698665ec49b1bc7.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78d21e209131cea7a9a2.md>)

Recorded evaluation configuration: {"Base Model":"SD3-M"}. This record does not assert independent reproduction.

### PG-ICLR’24

Density: 1.103. Result ID: 699b08845698665ec49b1bc4.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713f1e209131cea500fd.md>)

Recorded evaluation configuration: {"Base Model":"SD3-M"}. This record does not assert independent reproduction.

### PG-ICLR’24

Density: 1.039. Result ID: 699b08845698665ec49b1bbe.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713f1e209131cea500fd.md>)

Recorded evaluation configuration: {"Base Model":"SD2.1"}. This record does not assert independent reproduction.

### GASS

Density: 1.015. Result ID: 699b08845698665ec49b1bc2.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

Recorded evaluation configuration: {"Base Model":"SD2.1"}. This record does not assert independent reproduction.

### SPELL-ICML’25

Density: 0.912. Result ID: 699b08845698665ec49b1bc1.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78d21e209131cea7a9a2.md>)

Recorded evaluation configuration: {"Base Model":"SD2.1"}. This record does not assert independent reproduction.

### IG-NeurIPS’24

Density: 0.901. Result ID: 699b08845698665ec49b1bc0.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa758c1e209131cea6889d.md>)

Recorded evaluation configuration: {"Base Model":"SD2.1"}. This record does not assert independent reproduction.

### CADS-ICLR’24

Density: 0.897. Result ID: 699b08845698665ec49b1bbf.

[Reporting paper](<https://www.sota2.com/research/paper/6997ea7209c33900fcd76506.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73261e209131cea5ab3e.md>)

[Implementation](<https://github.com/CompVis/latent-diffusion>)

Recorded evaluation configuration: {"Base Model":"SD2.1"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
