HunyuanCustom Brings Single-Picture Video Deepfakes, With Audio and Lip Sync

This text discusses a brand new launch of a multimodal Hunyuan Video world mannequin referred to as ‘HunyuanCustom’. The brand new paper’s breadth of protection, mixed with a number of points in lots of the provided instance movies on the mission web page*, constrains us to extra basic protection than traditional, and to restricted copy of the large quantity of video materials accompanying this launch (since lots of the movies require vital re-editing and processing in an effort to enhance the readability of the structure).

Please be aware moreover that the paper refers back to the API-based generative system Kling as ‘Keling’. For readability, I confer with ‘Kling’ as a substitute all through.

 

Tencent is within the strategy of releasing a brand new model of its Hunyuan Video mannequin, titled HunyuanCustom. The brand new launch is seemingly able to making Hunyuan LoRA fashions redundant, by permitting the person to create ‘deepfake’-style video customization by means of a single picture:

Click on to play. Immediate: ‘A person is listening to music and cooking snail noodles within the kitchen’. The brand new technique in comparison with each close-source and open-source strategies, together with Kling, which is a major opponent on this area. Supply: https://hunyuancustom.github.io/ (warning: CPU/memory-intensive website!)

Within the left-most column of the video above, we see the one supply picture provided to HunyuanCustom, adopted by the brand new system’s interpretation of the immediate within the second column, subsequent to it. The remaining columns present the outcomes from numerous proprietary and FOSS programs: Kling; Vidu; Pika; Hailuo; and the Wan-based SkyReels-A2.

Within the video beneath, we see renders of three situations important to this launch: respectively, particular person + object; single-character emulation; and digital try-on (particular person + garments):

Click on to play. Three examples edited from the fabric on the supporting website for Hunyuan Video.

We are able to discover just a few issues from these examples, principally associated to the system counting on a single supply picture, as a substitute of a number of photographs of the identical topic.

Within the first clip, the person is actually nonetheless going through the digital camera. He dips his head down and sideways at not rather more than 20-25 levels of rotation, however, at an inclination in extra of that, the system would actually have to start out guessing what he seems like in profile. That is onerous, most likely inconceivable to gauge precisely from a sole frontal picture.

Within the second instance, we see that the little woman is smiling within the rendered video as she is within the single static supply picture. Once more, with this sole picture as reference, the HunyuanCustom must make a comparatively uninformed guess about what her ‘resting face’ seems like. Moreover, her face doesn’t deviate from camera-facing stance by greater than the prior instance (‘man consuming crisps’).

Within the final instance, we see that for the reason that supply materials – the lady and the garments she is prompted into carrying – are usually not full photographs, the render has cropped the state of affairs to suit – which is definitely relatively resolution to a knowledge situation!

The purpose is that although the brand new system can deal with a number of photographs (reminiscent of particular person + crisps, or particular person + garments), it doesn’t apparently permit for a number of angles or various views of a single character, in order that numerous expressions or uncommon angles may very well be accommodated. To this extent, the system might due to this fact battle to switch the rising ecosystem of LoRA fashions which have sprung up round HunyuanVideo since its launch final December, since these can assist HunyuanVideo to provide constant characters from any angle and with any facial features represented within the coaching dataset (20-60 photographs is typical).

Wired for Sound

For audio, HunyuanCustom leverages the LatentSync system (notoriously onerous for hobbyists to arrange and get good outcomes from) for acquiring lip actions which can be matched to audio and textual content that the person provides:

Options audio. Click on to play. Numerous examples of lip-sync from the HunyuanCustom supplementary website, edited collectively.

On the time of writing, there aren’t any English-language examples, however these seem like relatively good – the extra so if the tactic of making them is easily-installable and accessible.

Modifying Present Video

The brand new system provides what seem like very spectacular outcomes for video-to-video (V2V, or Vid2Vid) enhancing, whereby a section of an present (actual) video is masked off and intelligently changed by a topic given in a single reference picture. Beneath is an instance from the supplementary supplies website:

Click on to play. Solely the central object is focused, however what stays round it additionally will get altered in a HunyuanCustom vid2vid move.

As we are able to see, and as is commonplace in a vid2vid state of affairs, the complete video is to some extent altered by the method, although most altered within the focused area, i.e., the plush toy. Presumably pipelines may very well be developed to create such transformations beneath a rubbish matte method that leaves nearly all of the video content material similar to the unique. That is what Adobe Firefly does beneath the hood, and does fairly properly –  however it’s an under-studied course of within the FOSS generative scene.

That stated, a lot of the various examples offered do a greater job of concentrating on these integrations, as we are able to see within the assembled compilation beneath:

Click on to play. Various examples of interjected content material utilizing vid2vid in HunyuanCustom, exhibiting notable respect for the untargeted materials.

A New Begin?

This initiative is a growth of the Hunyuan Video mission, not a tough pivot away from that growth stream. The mission’s enhancements are launched as discrete architectural insertions relatively than sweeping structural modifications, aiming to permit the mannequin to take care of id constancy throughout frames with out counting on subject-specific fine-tuning, as with LoRA or textual inversion approaches.

To be clear, due to this fact, HunyuanCustom is just not skilled from scratch, however relatively is a fine-tuning of the December 2024 HunyuanVideo basis mannequin.

Those that have developed HunyuanVideo LoRAs might surprise if they may nonetheless work with this new version, or whether or not they should reinvent the LoRA wheel but once more if they need extra customization capabilities than are constructed into this new launch.

Generally, a closely fine-tuned launch of a hyperscale mannequin alters the mannequin weights sufficient that LoRAs made for the sooner mannequin is not going to work correctly, or in any respect, with the newly-refined mannequin.

Generally, nonetheless, a fine-tune’s recognition can problem its origins: one instance of a fine-tune changing into an efficient fork, with a devoted ecosystem and followers of its personal, is the Pony Diffusion tuning of Secure Diffusion XL (SDXL). Pony at present has 592,000+ downloads on the ever-changing CivitAI area, with an enormous vary of LoRAs which have used Pony (and never SDXL) as the bottom mannequin, and which require Pony at inference time.

Releasing

The mission web page for the brand new paper (which is titled HunyuanCustom: A Multimodal-Pushed Structure for Custom-made Video Era) options hyperlinks to a GitHub website that, as I write, simply grew to become purposeful, and seems to include all code and needed weights for native implementation, along with a proposed timeline (the place the one vital factor but to come back is ComfyUI integration).

On the time of writing, the mission’s Hugging Face presence remains to be a 404. There may be, nonetheless, an API-based model of the place one can apparently demo the system, as long as you may present a WeChat scan code.

I’ve not often seen such an elaborate and intensive utilization of such all kinds of tasks in a single meeting, as is clear in HunyuanCustom – and presumably a few of the licenses would in any case oblige a full launch.

Two fashions are introduced on the GitHub web page: a 720px1280px model requiring 8)GB of GPU Peak Reminiscence, and a 512px896px model requiring 60GB of GPU Peak Reminiscence.

The repository states ‘The minimal GPU reminiscence required is 24GB for 720px1280px129f however very gradual…We advocate utilizing a GPU with 80GB of reminiscence for higher technology high quality’ – and iterates that the system has solely been examined to date on Linux.

The sooner Hunyuan Video mannequin has, since official launch, been quantized right down to sizes the place it may be run on lower than 24GB of VRAM, and it appears cheap to imagine that the brand new mannequin will likewise be tailored into extra consumer-friendly varieties by the group, and that it’s going to rapidly be tailored to be used on Home windows programs too.

As a consequence of time constraints and the overwhelming quantity of data accompanying this launch, we are able to solely take a broader, relatively than in-depth take a look at this launch. Nonetheless, let’s pop the hood on HunyuanCustom a bit.

A Have a look at the Paper

The information pipeline for HunyuanCustom, apparently compliant with the GDPR framework, incorporates each synthesized and open-source video datasets, together with OpenHumanVid, with eight core classes represented: people, animals, crops, landscapes, autos, objects, structure, and anime.

Preliminary filtering begins with PySceneDetect, which segments movies into single-shot clips. TextBPN-Plus-Plus is then used to take away movies containing extreme on-screen textual content, subtitles, watermarks, or logos.

To deal with inconsistencies in decision and length, clips are standardized to 5 seconds in size and resized to 512 or 720 pixels on the brief facet. Aesthetic filtering is dealt with utilizing Koala-36M, with a customized threshold of 0.06 utilized for the customized dataset curated by the brand new paper’s researchers.

The topic extraction course of combines the Qwen7B Giant Language Mannequin (LLM), the YOLO11X object recognition framework, and the favored InsightFace structure, to establish and validate human identities.

For non-human topics, QwenVL and Grounded SAM 2 are used to extract related bounding bins, that are discarded if too small.

Multi-subject extraction makes use of Florence2 for bounding field annotation, and Grounded SAM 2 for segmentation, adopted by clustering and temporal segmentation of coaching frames.

The processed clips are additional enhanced by way of annotation, utilizing a proprietary structured-labeling system developed by the Hunyuan crew, and which furnishes layered metadata reminiscent of descriptions and digital camera movement cues.

Masks augmentation methods, together with conversion to bounding bins, have been utilized throughout coaching to scale back overfitting and make sure the mannequin adapts to numerous object shapes.

Audio knowledge was synchronized utilizing the aforementioned LatentSync, and clips discarded if synchronization scores fall beneath a minimal threshold.

The blind picture high quality evaluation framework HyperIQA was used to exclude movies scoring beneath 40 (on HyperIQA’s bespoke scale). Legitimate audio tracks have been then processed with Whisper to extract options for downstream duties.

The authors incorporate the LLaVA language assistant mannequin through the annotation section, and so they emphasize the central place that this framework has in HunyuanCustom. LLaVA is used to generate picture captions and help in aligning visible content material with textual content prompts, supporting the development of a coherent coaching sign throughout modalities:

By leveraging LLaVA’s vision-language alignment capabilities, the pipeline beneficial properties a further layer of semantic consistency between visible components and their textual descriptions – particularly useful in multi-subject or complex-scene situations.

Customized Video

To permit video technology based mostly on a reference picture and a immediate, the 2 modules centered round LLaVA have been created, first adapting the enter construction of HunyuanVideo in order that it might settle for a picture together with textual content.

This concerned formatting the immediate in a means that embeds the picture immediately or tags it with a brief id description. A separator token was used to cease the picture embedding from overwhelming the immediate content material.

Since LLaVA’s visible encoder tends to compress or discard fine-grained spatial particulars through the alignment of picture and textual content options (notably when translating a single reference picture right into a basic semantic embedding), an id enhancement module was included. Since almost all video latent diffusion fashions have some issue sustaining an id with out an LoRA, even in a five-second clip, the efficiency of this module in group testing might show vital.

In any case, the reference picture is then resized and encoded utilizing the causal 3D-VAE from the unique HunyuanVideo mannequin, and its latent inserted into the video latent throughout the temporal axis, with a spatial offset utilized to forestall the picture from being immediately reproduced within the output, whereas nonetheless guiding technology.

The mannequin was skilled utilizing Movement Matching, with noise samples drawn from a logit-normal distribution – and the community was skilled to recuperate the right video from these noisy latents. LLaVA and the video generator have been each fine-tuned collectively in order that the picture and immediate might information the output extra fluently and preserve the topic id constant.

For multi-subject prompts, every image-text pair was embedded individually and assigned a definite temporal place, permitting identities to be distinguished, and supporting the technology of scenes involving a number of interacting topics.

Sound and Imaginative and prescient

HunyuanCustom circumstances audio/speech technology utilizing each user-input audio and a textual content immediate, permitting characters to talk inside scenes that mirror the described setting.

To help this, an Identification-disentangled AudioNet module introduces audio options with out disrupting the id indicators embedded from the reference picture and immediate. These options are aligned with the compressed video timeline, divided into frame-level segments, and injected utilizing a spatial cross-attention mechanism that retains every body remoted, preserving topic consistency and avoiding temporal interference.

A second temporal injection module offers finer management over timing and movement, working in tandem with AudioNet, mapping audio options to particular areas of the latent sequence, and utilizing a Multi-Layer Perceptron (MLP) to transform them into token-wise movement offsets. This enables gestures and facial motion to comply with the rhythm and emphasis of the spoken enter with higher precision.

HunyuanCustom permits topics in present movies to be edited immediately, changing or inserting folks or objects right into a scene without having to rebuild your complete clip from scratch. This makes it helpful for duties that contain altering look or movement in a focused means.

Click on to play. An additional instance from the supplementary website.

To facilitate environment friendly subject-replacement in present movies, the brand new system avoids the resource-intensive method of current strategies such because the currently-popular VACE, or people who merge complete video sequences collectively, favoring as a substitute the compression  of a reference video utilizing the pretrained causal 3D-VAE –  aligning it with the technology pipeline’s inner video latents, after which including the 2 collectively. This retains the method comparatively light-weight, whereas nonetheless permitting exterior video content material to information the output.

A small neural community handles the alignment between the clear enter video and the noisy latents utilized in technology. The system exams two methods of injecting this data: merging the 2 units of options earlier than compressing them once more; and including the options body by body. The second technique works higher, the authors discovered, and avoids high quality loss whereas retaining the computational load unchanged.

Knowledge and Checks

In exams, the metrics used have been: the id consistency module in ArcFace, which extracts facial embeddings from each the reference picture and every body of the generated video, after which calculates the typical cosine similarity between them; topic similarity, by way of sending YOLO11x segments to Dino 2 for comparability; CLIP-B, text-video alignment, which measures similarity between the immediate and the generated video; CLIP-B once more, to calculate similarity between every body and each its neighboring frames and the primary body, in addition to temporal consistency; and dynamic diploma, as outlined by VBench.

As indicated earlier, the baseline closed supply rivals have been Hailuo; Vidu 2.0; Kling (1.6); and Pika. The competing FOSS frameworks have been VACE and SkyReels-A2.

Of those outcomes, the authors state:

‘Our [HunyuanCustom] achieves the most effective ID consistency and topic consistency. It additionally achieves comparable ends in immediate following and temporal consistency. [Hailuo] has the most effective clip rating as a result of it will possibly comply with textual content directions properly with solely ID consistency, sacrificing the consistency of non-human topics (the worst DINO-Sim). By way of Dynamic-degree, [Vidu] and [VACE] carry out poorly, which can be because of the small measurement of the mannequin.’

Although the mission website is saturated with comparability movies (the structure of which appears to have been designed for web site aesthetics relatively than simple comparability), it doesn’t at present characteristic a video equal of the static outcomes crammed collectively within the PDF, in regard to the preliminary qualitative exams. Although I embrace it right here, I encourage the reader to make a detailed examination of the movies on the mission website, as they provide a greater impression of the outcomes:

The authors remark right here:

‘It may be seen that [Vidu], [Skyreels A2] and our technique obtain comparatively good ends in immediate alignment and topic consistency, however our video high quality is healthier than Vidu and Skyreels, due to the great video technology efficiency of our base mannequin, i.e., [Hunyuanvideo-13B].

‘Amongst industrial merchandise, though [Kling] has video high quality, the primary body of the video has a copy-paste [problem], and generally the topic strikes too quick and [blurs], main a poor viewing expertise.’

The authors additional remark that Pika performs poorly by way of temporal consistency, introducing subtitle artifacts (results from poor knowledge curation, the place textual content components in video clips have been allowed to pollute the core ideas).

Hailuo maintains facial id, they state, however fails to protect full-body consistency. Amongst open-source strategies, VACE, the researchers assert, is unable to take care of id consistency, whereas they contend that HunyuanCustom produces movies with robust id preservation, whereas retaining high quality and variety.

Subsequent, exams have been performed for multi-subject video customization, towards the identical contenders. As within the earlier instance, the flattened PDF outcomes are usually not print equivalents of movies obtainable on the mission website, however are distinctive among the many outcomes offered:

The paper states:

‘[Pika] can generate the desired topics however displays instability in video frames, with cases of a person disappearing in a single state of affairs and a lady failing to open a door as prompted. [Vidu] and [VACE] partially seize human id however lose vital particulars of non-human objects, indicating a limitation in representing non-human topics.

‘[SkyReels A2] experiences extreme body instability, with noticeable modifications in chips and quite a few artifacts in the suitable state of affairs.

‘In distinction, our HunyuanCustom successfully captures each human and non-human topic identities, generates movies that adhere to the given prompts, and maintains excessive visible high quality and stability.’

An additional experiment was ‘digital human commercial’, whereby the frameworks have been tasked to combine a product with an individual:

For this spherical, the authors state:

‘The [results] display that HunyuanCustom successfully maintains the id of the human whereas preserving the main points of the goal product, together with the textual content on it.

‘Moreover, the interplay between the human and the product seems pure, and the video adheres carefully to the given immediate, highlighting the substantial potential of HunyuanCustom in producing commercial movies.’

One space the place video outcomes would have been very helpful was the qualitative spherical for audio-driven topic customization, the place the character speaks the corresponding audio from a text-described scene and posture.

The authors assert:

‘Earlier audio-driven human animation strategies enter a human picture and an audio, the place the human posture, apparel, and surroundings stay in keeping with the given picture and can’t generate movies in different gesture and surroundings, which can [restrict] their utility.

‘…[Our] HunyuanCustom permits audio-driven human customization, the place the character speaks the corresponding audio in a text-described scene and posture, permitting for extra versatile and controllable audio-driven human animation.’

Additional exams (please see PDF for all particulars) included a spherical pitting the brand new system towards VACE and Kling 1.6 for video topic alternative:

Of those, the final exams offered within the new paper, the researchers opine:

‘VACE suffers from boundary artifacts because of strict adherence to the enter masks, leading to unnatural topic shapes and disrupted movement continuity. [Kling], in distinction, displays a copy-paste impact, the place topics are immediately overlaid onto the video, resulting in poor integration with the background.

‘Compared, HunyuanCustom successfully avoids boundary artifacts, achieves seamless integration with the video background, and maintains robust id preservation—demonstrating its superior efficiency in video enhancing duties.’

Conclusion

It is a fascinating launch, not least as a result of it addresses one thing that the ever-discontent hobbyist scene has been complaining about extra these days – the shortage of lip-sync, in order that the elevated realism succesful in programs reminiscent of Hunyuan Video and Wan 2.1 may be given a brand new dimension of authenticity.

Although the structure of almost all of the comparative video examples on the mission website makes it relatively troublesome to match HunyuanCustom’s capabilities towards prior contenders, it should be famous that very, only a few tasks within the video synthesis area have the braveness to pit themselves in exams towards Kling, the industrial video diffusion API which is at all times hovering at or close to the highest of the leader-boards; Tencent seems to have made headway towards this incumbent in a relatively spectacular method.

 

* The difficulty being that a few of the movies are so vast, brief, and high-resolution that they won’t play in commonplace video gamers reminiscent of VLC or Home windows Media Participant, displaying black screens.

First revealed Thursday, Could 8, 2025