NextGAN: Towards Simple and Scalable GAN Training for Image and Video Generation

Xinhao Lou; Dongye Xie; Peng Sun; Chenyi Zhuang; Kaicheng Yu; Tao Lin

cs.CVNov 4, 2026

NextGAN: Towards Simple and Scalable GAN Training for Image and Video Generation

Authors: Xinhao Lou, Dongye Xie, Peng Sun, Chenyi Zhuang, Kaicheng Yu, Tao Lin

Organizations: Westlake University · University of Cambridge · Zhejiang University · Independent Researcher

Abstract

Adversarial training enables one- and few-step image and video generation, but conventional adversarial losses supervise critic values without directly specifying the gradients used to update the generator. We present NEXTGAN, which trains a scalar critic by regressing its input gradient onto displacements between independently paired reference and generated features at interpolated locations in frozen representation spaces. For a fixed generator, this regression is equivalent to an adversarial objective with a gradient-energy penalty, while the generator retains the standard potential-based GAN update. The same objective adapts pre-trained image and video generators with up to 14B parameters, without auxiliary losses, online teacher scores, or trajectories. On ImageNet $256 \times 256$ , NEXTGAN achieves one-step FIDs of 2.87 with pixel-space JiT-B and 1.54 with latent-space DiT-XL/1; its adversarial form further reaches 1.36. For text-to-video generation, a single trained Wan2.1-1.3B model supports one-, two-, and four-step sampling, achieving the highest VBench total scores among the compared 1.3B methods at one and two steps. At two steps, NEXTGAN scores 83.13, exceeding rCM and AnyFlow by 1.25 and 0.47 points, respectively. Code is available at https://github.com/LINs-lab/NextGAN.git.

Figures & tables

Figure 1: Qualitative samples from NextGAN . Left: 1-NFE class-conditional samples on ImageNet $256\times 256$ . Right: four frames from each of four generated videos.

Figure 2: How NextGAN learns to guide the generator. $\mathbf{x}_{\mathrm{p}}$ and $\mathbf{x}_{\mathrm{r}}$ are generated and real features; $\mathbf{x}_{\alpha}$ lies between them. First, the critic loss $\mathcal{L}_{D}$ trains $\nabla_{\mathbf{x}}u_{\boldsymbol{\phi}}$ to predict their displacement, with endpoints fixed ( $\operatorname{sg}$ : stop gradient). Then, the generator loss $\mathcal{L}_{G}$ moves generated features uphill in the fixed potential $u_{\boldsymbol{\phi}}$ . The surface depicts $-u_{\boldsymbol{\phi}}$ , so this motion appears downhill. These two updates alternate throughout training.

Figure 3: Mode coverage during training. Rows: NextGAN and GAN. Columns: samples from fixed latent probes at selected iterations and the target distribution. The curves below each row report mode imbalance,

8-N_{\mathrm{eff}}

; lower is better.

Method	Space	Backbone	Params	NFE	FID $\downarrow$	IS $\uparrow$
Consistency / flow-map distillation
Shortcut ( Frans et al., 2025 )	latent	DiT-XL/2	676M	1	10.60	–
PaGoDA ( Kim et al., 2024 )	pixel	ADM	301M	1	1.56	259.61
Diffusion / flow matching
MeanFlow ( Geng et al., 2025 )	latent	XL/2	676M	1	3.43	–
ADM ( Dhariwal & Nichol, 2021 )	pixel	ADM-G + ADM-U	866M	$250{+}250$	3.94	215.8

Table 1: Class-conditional ImageNet-256 results. Baselines are quoted from their original papers and may use different evaluators. NextGAN Params denote nominal backbone sizes (implementation counts: App. A ). NFE counts sequential generator evaluations, including guidance and autoregressive inner steps where applicable.

Model	Online teacher	NFE	Evaluation Scores $\uparrow$
			Quality	Semantic	Total
1.3B parameters
Wan2.1 T2V ( Wan Team, 2025 )	–	$50\times 2$	82.91	80.22	82.37
rCM ( Zheng et al., 2025 )	yes	1	81.76	76.34	80.68
AnyFlow ( Gu et al., 2026 )	yes	1	78.74	75.63	78.12
Wan2.1 $+$ DMD2 ( Yin et al., 2024 )	yes	1	80.30	71.85	78.61

Table 2: Text-to-video generation on VBench at

832\times 480\times 81

. Results are grouped by model size and NFE. “Online teacher” indicates whether training requires online teacher scores or trajectories. Higher is better for all three scores.

Panel of Figure 4 — Figure 4: Video generation with GAN, WGAN, TBSM (GAN), and NextGAN . Left: NextGAN achieves the highest quality, semantic, and total VBench scores (best in bold). Right: NextGAN shows clearer subject identity, more coherent object structure, and more accurate text generation.

Panel of Figure 5 — Figure 5: Six manually selected prompts (columns) with manually selected frames at matching indices across methods. Rows: rCM, AnyFlow, and NextGAN (all 1.3B).

Figure 6: Intra-prompt sample diversity under four-step sampling (Wan2.1-1.3B). Left: Wan2.1 $+$ DMD2; right: NextGAN . Rows show fish, flowers, and man; each panel shows five seeds per prompt, with matched seed indices across methods. Corresponding videos from both methods are shown at the same frame index.

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Panel of Figure 7 — Figure 7: Training objective and feature space. (a) JiT-B training curves through 50K updates (logarithmic FID-50K axis). (b) Frozen representer ablations on JiT-B at 20K updates. Lower is better.

Generative Adversarial Networks

Generative Adversarial Networks (GANs) train competing generators and discriminators to synthesize data, translate images, and model complex distributions.

See all 182 papers

AAD-1: Asymmetric Adversarial Distillation for One-Step Autoregressive Video Generation
Jun 2, 2026Haobo Li, Yanhong Zeng, Yunhong Lu +6Image-to-Video Generation Autoregressive Video Generation
AutoLab, SAI, SJTU · Ant Group · Department of Automation, Tsinghua University +2
Continuous Adversarial Flow Models
Apr 13, 2026Shanchuan Lin, Ceyuan Yang, Zhijie Lin +2Flow-Based Generative Modeling Image Generation
ByteDance Seed
Direct Forcing: Single-Stage Training with Representation-Space Supervision for Few-Step Autoregressive Video Generation
Oct 22, 2026Haojun Xu, Jie Huang, Chuyang Zhao +6T2V Generation Autoregressive Video Generation
Beihang University · JD Future Academy

Jun 2, 2026J/K move · Enter open · S save

AAD-1: Asymmetric Adversarial Distillation for One-Step Autoregressive Video Generation

Haobo Li, Yanhong Zeng, Yunhong Lu, Jiapeng Zhu+5

AutoLab, SAI, SJTU · Ant Group · Department of Automation, Tsinghua University+2

Image-to-Video Generation Autoregressive Video Generation Knowledge Distillation Temporal Consistency in Video Generation Adversarial Knowledge Distillation Single-Step Video Generation

Open paper AskPDF arXiv

We present AAD-1, an Asymmetric Adversarial Distillation framework for One-step autoregressive image-to-video generation. State-of-the-art methods adopt adversarial distillation but suffer from motion collapse and training instability, resulting in static videos. AAD-1 addresses these challenges through two key designs in architecture and training strategy. Our key architectural insight is to break the symmetry between generator and discriminator. While the generator remains causal to preserve autoregressive sampling capability, the discriminator attends bidirectionally over the full spatiotemporal context and produces a single holistic realism score for the entire video sequence. This asymmetric design enables the discriminator to effectively detect global temporal failures and long-range drift that cause motion collapse in autoregressive generation. To stabilize training, we introduce a phased strategy that first uses distribution matching to bootstrap a stable one-step generator, providing a warm-up phase that brings the student distribution closer to the teacher before adversarial distillation begins. Extensive experiments on VBench demonstrate that AAD-1 achieves state-of-the-art performance in one-step autoregressive video generation.

Abstract

Generative Adversarial Networks

Explore similar work

AAD-1: Asymmetric Adversarial Distillation for One-Step Autoregressive Video Generation

Continuous Adversarial Flow Models

Direct Forcing: Single-Stage Training with Representation-Space Supervision for Few-Step Autoregressive Video Generation