I need a video upscaler that de-noises old footage and does not create "wax face" smoothing artifacts.

If your priority is removing noise while preserving natural skin texture, you generally want a tool that preserves detail rather than one that tries to invent it. The "wax face" look usually comes from combining aggressive denoising with AI face reconstruction.

Here are the options I'd recommend, in order:

1. Topaz Video AI (best overall)

This is still the strongest choice for old VHS, Hi8, MiniDV, DVD, and archival footage—as long as you avoid the automatic presets. Users consistently report that manual tuning produces much more natural results than the default enhancement settings.

To avoid waxy faces:

• Use Proteus (Manual) rather than Auto.
• Use Nyx (or the dedicated denoise model) only when the footage is genuinely noisy.
• Keep denoise and sharpening conservative.
• Disable Face Recovery unless faces are severely damaged.
• Upscale only 2× if possible instead of jumping directly to 4K.
• If the image looks too "clinical," add a subtle amount of film grain afterward.

2. Neat Video + Topaz (my favorite workflow)

If you don't mind two steps, this usually gives the most natural results:

1. Denoise lightly in Neat Video
2. Upscale in Topaz Video AI
3. Finish with a tiny amount of grain

Neat Video removes actual sensor and compression noise rather than trying to recreate facial features, which helps preserve pores, wrinkles, and fabric texture. Many restoration professionals and experienced users prefer this split workflow.

3. DaVinci Resolve Studio

If realism matters more than maximum sharpness, Resolve is excellent.

Its temporal and spatial noise reduction plus Super Scale are much more conservative than many AI upscalers:

• preserves skin texture well
• keeps natural film grain
• avoids hallucinated facial detail

The tradeoff is that it won't "rescue" extremely poor footage as dramatically as Topaz.

4. AVCLabs Video Enhancer AI

A good alternative if you find Topaz too aggressive. It generally errs on the side of preserving the original image rather than aggressively reconstructing it, though it usually extracts a little less detail than Topaz.

What causes "wax face"

Regardless of software, these settings are usually responsible:

• Face enhancement/recovery enabled
• Denoise pushed too high
• Strong sharpening after denoising
• Trying to upscale 480p straight to 4K
• Attempting to remove every bit of grain

Leaving a little grain almost always looks more convincing than perfectly smooth skin.

My recommendations by source

If you tell me:

• the source format (VHS, Hi8, MiniDV, DVD, 8mm film, etc.),
• the resolution, and
• whether you have an NVIDIA GPU (and which model),

I can suggest settings that are much less likely to produce smoothing artifacts.

The "wax face" (or plastic/claymation) effect happens when an AI model tries to aggressively hallucinate or smooth out skin textures where the source footage is too degraded to provide real data . Avoiding this requires an upscaler that allows manual slider control and a multi-stage workflow (de-noising first at original resolution, then sharpening, then scaling) rather than relying on a single "magic" auto button.

The top software options capable of avoiding the waxy look on old footage include:

1. Topaz Video (with Starlight or Proteus/Nyx models)
   • Best for: Professional and archival restoration.
   • Why it works without wax faces: Topaz features specialized models (like Nyx for precise noise reduction and Proteus or Starlight for structural fidelity). Crucially, you must switch the settings from Auto to Manual and keep the noise reduction and sharpening sliders below 30-40% . Cranking them up is what causes the plastic face artifact.
   • Learn more and check system specs on the official Topaz Labs site.
2. Neat Video (Plugin for Premiere, DaVinci Resolve, or Final Cut)
   • Best for: Pure, artifact-free analog de-noising before upscaling.
   • Why it works without wax faces: Neat Video isn't a generative AI upscaler; it’s an advanced profile-based digital filter. It targets actual camera sensor grain, tape hiss, and digital noise profiles with extreme mathematical precision without "inventing" pixels or warping facial structures.
   • Pro-Tip Workflow: Run your footage through Neat Video first to clean the noise organically, then take that clean file into an upscaler for the resolution bump. Check out user implementation details via the Neat Video Plugin page.
3. Aiarty Video Enhancer
   • Best for: A more accessible local workflow with built-in strength limiters.
   • Why it works without wax faces: It provides dedicated strength control sliders explicitly designed to prevent the over-sharpened, plastic look. By utilizing lower settings on its natural restoration models, it preserves underlying skin texture and natural film grain rather than smearing them into a flat paste.
   • Read more about its processing modes on Aiarty Video Enhancer.

The Golden Rule to Avoid Waxy Faces

Regardless of which software you use, do not upscale and de-noise in a single 4x jump pass.

• Pass 1: Run a mild noise-reduction/de-interlace pass at 100% original resolution.
• Pass 2: Export that intermediate file, then run a conservative structural enhancement/sharpening pass (sliders low).
• Pass 3: Upscale the resolution last.
• Always render a 2-to-3-second tight crop of a face before committing to a multi-hour render.

If you'd like, tell me:

• What is the source format of your footage (e.g., 8mm film, VHS, MiniDV, 90s digital)?
• What hardware/GPU are you running on your computer?

I can give you a step-by-step slider configuration to prevent artifacts for your specific material.

Avoiding the dreaded "wax face" or plastic doll filter requires stepping away from fully automated AI presets. When an AI upscale tool tries to aggressively guess missing skin pores or facial topology on low-resolution source files, it completely smooths away natural textures.

The industry benchmark for handling this balance is Topaz Video (utilizing its dedicated restoration and upscaling software suite), though open-source local solutions like Upscayl or node-based workflows can work if tuned tightly.

Watch this overview of Topaz Video settings and models to see how different parameters handle delicate facial textures and noise reduction:

2:54 Topaz AI Secret Settings 2026 🔥 Get Sharp 4K Quality ...56K views  · 23 days ago YouTube · Nihaal Edits

How to Configure Settings to Avoid Waxy Faces

If you are using a professional desktop suite like Topaz Video, adhere to this configuration strategy to preserve natural skin and grain:

1. Switch from Auto to Manual Control: Never trust the "Auto" or default settings for old, grainy, or low-res human-centric footage; Auto pushes detail recovery too high and triggers the plastic look.
2. Choose the Right Model:
   • Use Iris LQ (Low Quality) rather than MQ/HQ variants for heavily compressed or low-res faces, as higher models tend to hallucinate synthetic skin features on soft SD sources.
   • Alternatively, use Proteus (Manual) if you want general artifact removal without face-specific generation.
3. Dial Down the Aggressive Sliders:
   • Keep Improve Detail and Fix Compression low (set them anywhere from 5 to 20, or turn them completely off depending on the source).
   • Keep Recover Detail at a higher, stable balance (around 75 to 80) so the model relies on the original frame data instead of painting over it.
4. Reintroduce Subtle Film Grain: Add a tiny touch of grain back into the image (Amount: 1-2, Size: 1) during the final export pass. True organic grain tricks the eye into ignoring any digital smoothing and stops skin from looking rubbery.
5. Cap Your Upscale Factor: Don't jump straight from 480p to 4K in a single pass. Do a clean 2× scale first (or keep it at original resolution for a de-noising pass), export to a lossless format like ProRes 422 HQ, and only scale further if the intermediate file looks natural.

If you can share what resolution/format your source footage is (e.g., 480i VHS, 8mm film, MiniDV) and your hardware setup , I can recommend the exact multi-stage workflow pipeline to use.

The "wax face" look is almost always caused by overly aggressive denoising and AI face reconstruction, not by upscaling itself. The best tools let you control those stages separately rather than forcing a one-click enhancement. Community experience consistently favors workflows that preserve some texture and grain instead of trying to eliminate every imperfection.

Here's how I'd rank the current options:

1. Topaz Video AI — Best overall

If you only want one application, this is still the strongest choice.

What helps avoid waxy faces:

• Use Proteus (Manual) rather than Auto.
• Use Nyx (or the dedicated denoise model) only when the source is genuinely noisy.
• Disable or minimize Face Recovery unless faces are severely damaged.
• Keep sharpening and detail recovery conservative.
• Upscale only 2× when practical instead of jumping straight to 4K.

2. DaVinci Resolve Studio — Most natural texture

If realism is more important than maximum "wow" factor, Resolve is excellent.

Its temporal and spatial noise reduction combined with Super Scale generally:

• preserves pores and wrinkles,
• keeps film grain looking natural,
• avoids hallucinated facial detail.

The tradeoff is that it won't invent detail as aggressively as Topaz.

3. Neat Video + Topaz Video AI — Best professional workflow

This is the workflow many restoration enthusiasts prefer:

1. Light denoise in Neat Video.
2. Upscale in Topaz.
3. Optionally add a subtle amount of grain afterward.

Separating denoising from upscaling usually produces more believable skin texture than asking one AI model to do everything.

4. Video2X — Best free option

If you want something open source, Video2X is worth considering. It is more manual to configure but tends to avoid aggressive facial beautification because it relies on conventional upscaling models rather than heavy face reconstruction.

General settings that reduce "wax face"

Regardless of software:

• Use the weakest denoise that removes distracting noise.
• Avoid automatic face enhancement.
• Leave a little film grain or sensor noise.
• Process a short test clip before committing to a full render.
• If the image starts looking "too clean," you're probably over-processing it.

My recommendations by source

If you tell me what you're restoring (VHS, Hi8, MiniDV, DVD, Super 8, 16mm, or early digital camera footage), I can suggest settings that are known to preserve texture while cleaning up noise.

The "wax face" look is usually caused by over-aggressive denoising, face enhancement, or AI hallucination, not by upscaling itself. The best tools let you control those stages independently instead of applying a one-click "enhance everything" pipeline.

Here's how I'd rank the current options:

My top recommendation

If quality matters most:

1. Neat Video — remove only the actual sensor/compression noise.
2. Topaz Video AI — upscale 2× using a manual model like Proteus.
3. Add a small amount of film grain afterward if the image starts looking overly "digital."

This split workflow consistently produces more natural skin than asking one AI model to denoise, sharpen, and upscale simultaneously.

If using Topaz Video AI

Avoid the defaults. A good starting point is:

• Use Proteus Manual rather than Auto.
• Disable Face Recovery unless faces are severely damaged.
• Keep denoise conservative.
• Keep sharpening low.
• Upscale 2× instead of jumping straight to 4K.
• Leave a little grain—trying to remove every speck is what creates plastic-looking skin.

If your priority is "don't invent details"

Choose DaVinci Resolve Studio.

Its temporal and spatial noise reduction plus Super Scale are generally more conservative than AI restoration tools. You won't get as much apparent sharpness, but you'll also see fewer fake pores, melted faces, and painted-looking skin.

I would avoid

Many consumer "AI enhancer" apps that advertise one-click 4K restoration tend to:

• smooth skin excessively,
• over-sharpen edges,
• hallucinate eyelashes and hair,
• remove natural film grain.

Those effects often look impressive in thumbnails but break down when viewed frame-by-frame or on a large display.

If you can tell me what the source footage is (VHS, Hi8, MiniDV, Betacam, DVD, 8mm film scan, 480p digital video, etc.) and what GPU you have (or whether you're on a Mac), I can suggest the restoration workflow and settings that are known to work best for that specific format.

To de-noise old footage while avoiding the "wax face" smoothing artifact (where faces look like plastic), you need a tool that supports specialized temporal denoising and texture-preserving models. The "wax face" look is usually a result of over-smoothing by AI models meant to reduce grain, which can be mitigated by keeping the original texture intact.

Here are the best options based on your requirements:

1. Topaz Video AI (Top Recommendation)
   • Why it fits: It is specifically designed to handle old, grainy footage.
   • Artifact Control: Instead of standard "upscaling," use models like Iris (v2) or Proteus for high-quality restoration. Proteus allows you to adjust parameters manually, allowing you to reduce noise without losing facial details.
   • Key Advantage: It focuses on retaining natural facial structures and skin textures rather than replacing them with AI-generated smooth skin.
2. DaVinci Resolve + Topaz Video AI/Plugins
   • Why it fits: It allows for a hybrid approach. You can use DaVinci’s temporal noise reduction to clean up the motion, and then use AI models for the upscale.
   • Artifact Control: The key is to not over-denoise in the first step.
3. HitPaw Video Enhancer
   • Why it fits: Offers specific models for face restoration and general old footage enhancement.
   • Artifact Control: Good for removing noise while upscaling to 4K, often with better control over the "plastic" look than cheaper alternatives.

Tips to Avoid "Wax Face" Artifacts:

• Use Fine-Tune Models: Select models that prioritize detail retention over simple smoothness.
• Balance Noise Removal: If you use too much denoising, you will lose the pores and natural details, creating a "wax face".
• Lower the Upscale Factor: Sometimes, upscaling to 200% instead of 400% preserves more natural texture.

To help me narrow down the best tool, tell me:

• What is the source resolution (e.g., VHS, SD 480p, 16mm film)?
• Are you looking for a desktop software or an online tool?

If your priority is preserving natural facial texture rather than producing an aggressively "enhanced" look, the best tools are the ones that let you control denoising, sharpening, and upscaling independently. The "wax face" effect is usually caused by excessive temporal denoising, AI face reconstruction, or over-sharpening—not by upscaling itself.

Here's how I'd rank the current options:

1. urlTopaz Video AIturn0search1 — Best overall

   • Excellent for VHS, Hi8, MiniDV, DVD, and old digital footage.
   • Use Proteus (Manual) or Nyx rather than fully automatic modes.
   • Keep denoise conservative.
   • Disable or minimize Face Recovery unless faces are severely damaged.
   • Upscale 2× instead of 4× whenever possible.
   • If the image becomes too "clean," add a tiny amount of film grain afterward.

2. DaVinci Resolve Studio

   • The most natural-looking option if you're willing to tweak settings.
   • Temporal + spatial noise reduction followed by Super Scale tends to preserve pores, wrinkles, and film grain better than many AI-only upscalers.
   • It invents less detail, but that also means fewer artificial-looking faces.

3. Neat Video

   • Still one of the gold standards for denoising.
   • It removes actual noise rather than trying to regenerate facial features.
   • Many restoration workflows use:
     1. Neat Video (light denoise)
     2. Topaz Video AI (upscale)
     3. Color grade and add subtle grain This separation of tasks is widely recommended because it reduces plastic-looking skin.

Workflow that usually gives the most natural results

Instead of asking one AI model to do everything:

• Deinterlace (if needed)
• Light denoise
• Fix exposure/color
• Upscale
• Very light sharpening
• Add a subtle amount of grain

Professionals and restoration enthusiasts generally report that this preserves texture much better than heavy one-pass enhancement.

Settings to avoid

If you start seeing these, back off:

• Skin with no pores
• Hair turning into painted clumps
• "Glass" eyes
• Teeth becoming unnaturally white or smooth
• Clothing fabric losing weave texture

Once those artifacts appear, increasing sharpening usually makes them worse rather than restoring real detail.

My recommendations by budget

If you tell me what your source is (VHS, Hi8, MiniDV, Super 8 film scan, DVD, LaserDisc, etc.) and what GPU you have, I can recommend specific settings that minimize smoothing while preserving facial texture.

For upscaling and denoising old footage while avoiding "wax face" smoothing artifacts, Topaz Video AI is widely considered the industry standard. It offers specialized models designed to preserve natural texture rather than just smoothing everything over.

Here are the best options for this specific need:

• Topaz Video AI: Features models like Gaia-CG (good for CGI/animations), Artemis (good for denoising/deinterlacing), and Proteus (high control over parameters). To avoid the "wax face" look, you must decrease the 'Reduce Noise' setting and increase 'Revert Compression' or use the 'Fine-Tune' model to retain skin pores and natural grain.
• DaVinci Resolve (Studio): Uses Neural Engine AI for upscaling (Super Scale) and possesses excellent temporal noise reduction tools that, when used properly, maintain detail better than cheap upscalers.
• HitPaw Video Enhancer: Offers a "General Model" or "Face Model" specifically designed to enhance facial details in old videos without turning them into plastic.

Tips for Avoiding "Wax Face" Artifacts:

• Don't Over-Denoise: Excessive noise reduction is the primary cause of the wax look. Keep reduction settings moderate to leave some film grain.
• Use Fine-Tune Models: In Topaz, select the Proteus model and manually adjust the "Reduce Noise" and "Sharpen" sliders rather than using auto-presets.
• Add Grain Back: If the software makes it too smooth, add a very fine layer of synthetic film grain back to the video to mask the smooth, artificial texture.

Would you like:

• A comparison of subscription vs. perpetual pricing?
• A guide on using ProRes for the highest quality export?
• A specific recommendation for Black & White vs. Color footage?

Let me know what type of footage you are working with!

To upscale old footage, remove noise, and avoid the "wax face" (over-smoothed, plastic) look, you need AI tools that allow for granular control over texture retention and grain reduction.

Here are the best, industry-leading solutions for this specific request:

1. Topaz Video AI (Best Overall for Realism)

Topaz Video AI is widely considered the best software for this because its models are designed to understand the difference between actual image details and noise.

• Why it prevents "Wax Face": It features specialized models like Iris (designed for faces) and Nyx (for intense noise reduction) that maintain natural skin texture and pores.
• Key Features: It offers "Proteus" or "Iris" models, allowing you to fine-tune noise reduction vs. sharpening. Instead of turning skin into plastic, it reconstructs textures based on surrounding, higher-quality frames.

2. DaVinci Resolve (Studio Version)

DaVinci Resolve uses a powerful AI tool called the Neural Engine to perform advanced spatial and temporal noise reduction.

• Why it prevents "Wax Face": It provides professional-grade tools that allow you to separate color noise from luminance noise. This means you can clean up the grainy color without blurring the underlying sharp edges of a face.

3. HitPaw Video Enhancer

HitPaw is a more automated, user-friendly alternative that still provides high-quality results.

• Why it prevents "Wax Face": It includes a specific "General Denoise" model designed to balance enhancement with noise removal.

Summary of Tips for Avoiding "Wax Face"

Regardless of the tool, to avoid the wax look, you must follow these rules when processing:

1. Do Not Over-Denoise: Aim to reduce noise, not eliminate it entirely. A tiny amount of organic film grain prevents the image from looking fake.
2. Add Subtle Grain: If the upscaler makes the face too smooth, add a very fine layer of digital film grain back onto the image to restore the illusion of texture.
3. Use Specialized Face Models: Always choose models that specifically say "Face Enhancement" or "Portrait," rather than general sharpening models which over-sharpen skin textures.

Are you looking for a desktop application for a powerful computer, or a cloud-based tool?

If you tell me your operating system (Windows/Mac) and your experience level (beginner/pro), I can tell you which of these is best for you.

If your priority is removing noise while preserving natural skin texture, the key is choosing software that lets you separate denoising from enhancement. The "wax face" effect usually comes from aggressive face reconstruction, sharpening, or AI detail hallucination rather than denoising itself.

Here's how the leading options compare:

My recommendations

1. Topaz Video AI (best quality, if configured carefully)

Despite its reputation for dramatic AI enhancement, it can produce very natural restorations if you avoid the aggressive defaults.

For old footage:

• Use the dedicated Denoise filter (Nyx High-Fidelity) first.
• Keep sharpening low.
• Disable Face Recovery unless faces are severely damaged.
• Upscale only 2× if possible instead of jumping straight to 4K.
• Leave a little film grain instead of trying to eliminate every speck of noise. Topaz's dedicated denoise model is designed specifically to remove grain without adding extra enhancement.

2. DaVinci Resolve Studio (most natural)

If you're restoring family videos or archival material, Resolve's temporal and spatial noise reduction combined with Super Scale generally preserves pores, wrinkles, and natural texture better than more aggressive AI restoration. The tradeoff is that it won't invent as much detail from very poor sources.

3. Neat Video + Topaz (professional workflow)

Many restoration professionals split the process:

1. Denoise with Neat Video.
2. Upscale with Topaz.
3. Add a subtle amount of grain back if the image looks too "clean."

This approach often avoids the plastic-looking skin that can happen when one AI model tries to denoise, sharpen, and reconstruct faces simultaneously. Community discussions frequently recommend this workflow for archival footage.

Settings that help avoid "wax face"

Regardless of the software:

• Reduce noise before increasing sharpness.
• Avoid face enhancement unless it's truly needed.
• Keep detail recovery conservative.
• Preserve some fine grain and texture.
• Upscale in smaller steps (e.g. 480p → 1080p instead of 480p → 4K).

These choices tend to preserve a more photographic look and reduce the over-smoothed, synthetic appearance associated with "wax face."

If you can tell me:

• your source (VHS, Hi8, MiniDV, DVD, early digital camera, etc.),
• resolution (480i, 576i, 720p, etc.),
• and whether you're on Windows, macOS, or Linux,

I can recommend settings that are well suited to that specific type of footage.