AIARTY Image Enhancer | Where It Fits… and Where It Doesn’t

What happens when an image is technically “lost”?

Blurry focus, visible noise, low resolution. Normally, that’s where the process ends.

In this video, I test Aiarty Image Enhancer in a real-world workflow to see if it can actually recover detail or if it just creates the illusion of it.

This is not a surface-level look. I go through multiple images, different AI models, and show where the software delivers subtle, usable improvements and where it completely falls apart.

You’ll see how it handles facial detail, texture, and more challenging compositions, along with how it behaves when pushed too far.

The takeaway is simple. This is not a fix-all tool. But in the right situations, it can take an image that’s already strong and refine it just enough to make a difference.

If you care about maintaining a natural look while exploring AI-assisted workflows, this is worth understanding before adding it into your process.

-

Learn more about Aiarty Image Enhancer: https://www.aiarty.com/ai-image-enhancer/?ttref=2604-aia-ytb-video-kapo-mty

Use code - KAIZEN25

Get up to 49% OFF depending on your plan:
• Lifetime license for $79 instead of $155
• 1-year license for $65 instead of $85

If you’ve been curious about AI image enhancement, this is a solid entry point.

-
🎥 Chapters:
00:00 - Introduction - AIARTY Image Enhancer
01:20 - What The Software Does
01:55 - Hardware Compatibility Check
02:30 - Restore Color Mode | Settings
03:32 - Clear Temp Files On Exit | Detailed User Guide
04:10 - Real Portfolio Image Tests
07:15 - Perfomance Issues Higher Upscaling
08:00 - Dual Monitor Issues
08:24 - Zooming Issues
08:55 - Denoising And How It Works
10:25 - Restore Color Button Issue Combined With Noise Reduction
11:34 - Sample Images Included Feel AI-Heavy
12:00 - Final Thoughts
-

— CONNECT & PARTNERS —
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
📺 YouTube: https://www.youtube.com/kaizenoriginal
📘 Facebook - https://www.facebook.com/Kaizenoriginal/
📸 Instagram: https://www.instagram.com/kaizenoriginal/
🧵 https://www.threads.com/@kaizenoriginal
🌐 Website / Bookings: https://www.kaizenoriginal.com
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
If You Care About Ownership | Then DxO Is For You.
https://www.youtube.com/watch?v=5K7HWec_Dhw&t=1s
-
DxO PhotoLab 9 & the DxO Ecosystem
Use “KAIZEN15” for 15% off
https://shop.dxo.com/en/
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Love The Look Of Film? Then This Is For You.
https://www.youtube.com/watch?v=7NOodAAPTrA
-
Color Precision – (Lightroom Profiles)
https://www.colorprecision.com/rxmtx9
-
Color Precision – (Capture One Profiles)
https://www.colorprecision.com/vdmnun
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Dehancer Analog Film Emulation | Photo & Video
Discount Code KAIZEN10 for 10% Off
https://www.dehancer.com/
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Trackclub Music & FX | Music & Sound That Elevates
https://trackclub.com/referral-landing?via=kaizenoriginal
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Zhiyun Stabilizers & Accessories | Stability Without Compromise
https://store.zhiyun-tech.com/?ref=xweltrib&utm_campaign=xweltrib&utm_medium=goaffpro&utm_source=affiliate
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Tether Tools PRO | Stay Connected. Stay In Flow
https://tethertools.com/?ref=232
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
SIRUI Tripods & Lenses | Precision built for real work
https://tidd.ly/3ZWAoZ7
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
The Mitakon 85mm 1.2 GF Mount | My All-Time Favorite Lens To This Day.
https://www.youtube.com/watch?v=s0FanBsOk8Y

Mitakon | ZY Optics offer some of the best mount options and pricing I’ve seen to date when it comes to high-quality lenses. And right now, ZY Optics have a great offer on the Mitakon Creator 35mm f/2, coming in at just $199. A fast, compact prime that delivers a distinctive look and solid build quality at an accessible price point. A strong option for both photography and video when you want character without unnecessary complexity.

And as I mentioned in the video, the bokeh on these lenses is something else, with a creamy, soft, almost otherworldly character. I genuinely can’t recommend it enough at this price.

Available mounts for the Mitakon Creator 35mm f/2:
Sony FE • Canon EF • Nikon F • Pentax K

Check it out here:
https://zyoptics.net/product/mitakon-creator-35mm-f-2/?ref=Kaizenphoto - Affiliate Link
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Aputure Backdrops | Clean Sets. Honest Light
https://tidd.ly/4caGPPG
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Clearly - Your Health is Clearly One
https://clearly.eu/KAIZENPHOTO
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Some links may be affiliate links. I only recommend tools I genuinely use in my own workflow.
If this video was useful, feel free to like & subscribe. New videos every week. (Mostly :P)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
#Photography #Filmmaking #VisualStorytelling #CreativeProcess #ContentCreator

---

Closed Caption:

I've taken a lot of photos that just didn't 
turn out the way I wanted them to. Blurry,
noisy, low resolution, and usually 
once an image gets to that point,
you just leave it alone. But tools 
like this are starting to change that.
So, I've been testing a software called 
Aiarty Image Enhancer from Digiarty,
and it's built around upscaling, denoising, 
and recovering detail from images that aren't
in great condition to begin with. And 
instead of just sharpening a photo,
it tries to rebuild detail using different AI 
models depending on what you're working with.
And at the same time, you've still got your 
standard adjustment controls like exposure,
contrast, white balance, and color, 
which are really the foundation of
any editing workflow. So it's not 
just about running an AI process
here. You can actually work on the image 
properly inside the software as well.
Now, normally I don't structure my videos like 
this, but there's a lot going on here. So I'm
just going to go through what I liked and the 
things that I didn't like, point by point,
just to give you a better understanding of what 
this software is and what it can and can't do.
So what this is really trying to do is go beyond 
just sharpening. Instead of just boosting edges,
it's trying to add back detail that wasn't 
really there to begin with. And that can make
a noticeable difference depending on the 
image. And having a few different models
to choose from, it really helps because 
not every image responds the same way.
The workflow is very simple. You bring 
in your image, you choose an AI model,
and then you process it. You don't 
need to dig through menus just to
get started. The settings are clean and 
easy to move through. There's no clutter,
and everything feels like it's 
where it's supposed to be.
Now, there's also a hardware compatibility check 
built in, which is a really useful touch. It shows
what your system can handle before you start 
pushing things too far. And so you've got all
of these standard sliders here as well, exposure, 
contrast, color, and that matters because those
are still the backbone of editing. The 
AI helps, but it doesn't replace that.
Since everything is built into one place, you 
don't need to jump into another program just
to finish your image. You can go from start 
to finish right here in the software. And
there's also a restore color mode, which 
I really, really like, which helps bring
things back if AI shifts tones too much. It's 
a small feature, but it solves a real problem.
So upscaling is surprisingly fast, and in 
some cases going from lower resolution to
something much higher happens almost 
instantly. You can also choose between
CPU and GPU processing depending on your 
setup, which gives you more control. And
there's also the tile system where you 
can adjust how the image is processed in
segments. Higher values use more VRAM, 
but it can also improve performance.
And when it comes to exporting, you've also 
got control over output DPI, which is useful if
you're doing any kind of print work. It goes up to 
1,000 DPI if you need it. And there's also an SDR
and HDR option, which can add more contrast and 
depth. It's not something I'd use all the time,
and you do need an HDR monitor to really take 
advantage of it, but it's there if you need it.
Now, here's another feature I really like. 
You can also clear temporary files when you
exit the software, which helps keep things clean 
over time. And another thing this company does
really well is they provide a detailed user guide 
along with extra test images and tutorials. It's
useful to go through, but at some point you'll 
just want to start using the software yourself.
And just to mention, this software has been 
featured on platforms like PetaPixel and
Fstoppers, and the company has been around since 
2006. So it's not something new or untested.
So up until this point, I've mainly been testing 
it with the images included in the software, but
to get a better idea of how it actually performs, 
I also tested it on some of my own images.
On the first image, using the more 
detailed model with two-pass processing,
it actually did a really nice job. It brought 
out detail in the eyebrows, eyelashes, iris,
and the lips, but kept everything subtle. 
It doesn't feel overprocessed. It just
feels like the image has been refined 
a bit, which is exactly what you want.
Now, moving on to the second image. The second 
image is where things start to fall apart a bit.
This one is more challenging. The subject is 
slightly turned. There's more negative space,
and the left side of the face is softer and 
a bit out of focus. So it's not a bad image,
just a more complex one. It's shot with 
intent. It's intended to be like this.
Now, when I turned on face restoration here, it 
completely broke the image. It didn't just miss
slightly. It made things look unnatural. The 
face started to look distorted and almost like
something out of a horror film. I tried different 
modes like rebuild, fidelity, and overlay options,
but none of them really worked. So in this 
case, face restoration just isn't usable.
So I turned that off and used the more 
detailed GAN V2 model with two-pass
processing just to see what we can do 
with this image here. And even there
it's struggling. I don't really know 
why. But instead of recovering detail,
it actually made parts of the image worse. The 
left eye didn't improve, and details in things
like the headdress and the earrings actually 
got reduced. The only thing it handled fairly
well here was the skin. It smoothed things out a 
bit, but overall this one was definitely a miss.
Now, the third image was intentionally softer 
with the focus on the earrings rather than the
face. The subject is more front-facing, so 
the software recognizes the face better. But
once you turn on face restoration, it still 
changes the image too much. It removes detail
from important elements. It flattens the skin 
tones and even starts altering facial features.
Turning it off and using V2 gives a 
slightly better result, but still not
really an improvement. And in fact, I'm getting 
more usable detail from the original image. Now,
to be fair, this is a more difficult image, 
but it's also the kind of situation where you'd
expect tools like this to help. So for this 
one, it just didn't quite get there either.
The last image is where it works really well. 
Again, with the eyes in focus and the image
already strong, both V2 and V3 bring out small 
details in a very subtle and controlled way.
Skin tones stay natural and nothing gets overdone, 
and everything just feels balanced. So this one I
really like. And this is where I can actually 
see myself using it. Not to fix a bad image,
but to push a good one just a little bit 
further, from a 9 out of 10 to a 10 out of 10.
Now, after going through these images, that's 
where a few limitations start to show up.
Talking about performance and responsiveness, 
one thing I did notice is how the sliders
behave at higher resolutions. Up to around 
4K, everything feels smooth and responsive,
but once you move up to 8K, there's a bit 
of a delay before the image updates. Now,
to be fair, working at 8K isn't something most 
people are doing all the time, but it does show
that as you push the resolution higher, you 
start to lose that real-time feedback a bit.
What's interesting though is that 
the upscaling itself is very fast,
but interacting with the image at that 
level slows things down slightly. So
it's not really a dealbreaker, 
just something to be aware of.
And here's another issue I ran into. So 
if you're running a dual monitor setup,
this is something to keep in mind. The software 
works fine when you open it on one screen, but if
you drag it between monitors, especially between 
one that's horizontal and one that's vertical,
it can break the interface. So you end 
up having to close it and reopen it on
the screen you want to use it on. It's not a huge 
issue, but it does interrupt your workflow a bit.
Zooming is also limited here. 
You can zoom in and out,
but only until the image hits the edges of 
the canvas. So you can't zoom out beyond
that point to get a wider overview like 
you can in something like Photoshop. Now
this is really important to me, and I hope 
this can be fixed with software updates.
Now looking at the different output qualities, the 
newer V3 model isn't always actually better than
the V2. And in some cases, the V2 gives a more 
natural result. So I ended up using that more.
So one major thing that's missing for me 
here, or at least it's not very clear,
is denoising. Denoising is a pretty 
important part of any editing workflow,
especially if you're shooting at higher ISO. 
And right now, it's not very obvious where that
sits in the software. There's no dedicated 
control for it in the main sliders, and I
do think this is something that should have its 
own section just to make it easier to work with.
Now, if you dig a bit deeper, there is some 
functionality tied to the AI models. There's
a model called Real Photo V3, which is supposed 
to handle things like deblurring and denoising,
but from my testing, it's not really doing that.
I tested this on an image shot at ISO 6400, which 
honestly isn't that extreme with modern cameras,
and instead of reducing noise, it actually 
introduced more noise, more artifacts,
and more color noise into the image. Even when 
switching between the models like GAN V2 and V3,
I'm still not seeing any proper noise 
reduction. There's no real smoothing,
no cleanup, nothing that you 
would expect from a noise feature.
And just to put that in perspective, if you 
compare it to something like DxO PureRAW,
which is very strong in this area, you can take 
extremely noisy images and make them usable again.
Here, that's just not happening right now. So 
this is definitely an area that needs more work,
both in terms of how it's presented in 
the software and how it actually performs.
But I wasn't quite ready to let this 
go yet because something just felt off.
So I took a second look at 
my image here of Radagast,
and I found the culprit. We have this box right 
here that restores color. When it's ticked in,
all these artifacts and color noise 
are introduced. If you untick that box,
the software actually does a pretty solid job 
of removing noise and bringing back detail.
But I feel that it is a little bit 
misleading because if I look at one
of the sample images that's included here, 
let's take a look at this tiger here. The
before image clearly shows that it's out 
of focus, really grainy, and it looks bad.
But if I use the more detailed GAN V2 
and I have the restore color ticked in,
it doesn't introduce any artifacts 
or noise like it does with Radagast.
Now, if I untick this, it gets 
more bleached. But if I tick it in,
it brings back those original 
orange colors of the tiger,
and there's no artifacting and no color noise. 
So what's really going on here? I don't know.
But if you're using the software and 
you want to get rid of that noise and
you're using one of your original images, 
make sure that this box is not ticked in.
Talking about content and exporting the images, 
the sample images included with the software
feel very AI-heavy and not very realistic. You 
can download better examples from their site,
but it would be nice to have 
those included from the start.
So overall, I think this is a pretty 
capable piece of software. But it's
also one of those tools where you start 
to notice that some things it does really,
really well and other things not so 
much. And that's okay because this
is clearly software that's trying to push 
boundaries and do something a bit different.
For my own use, I wouldn't rely on it for 
everything, but in the right situation,
it fits in really well. For me personally, 
this is more of a complementary tool. If I've
got an image that's already in a good place 
and I just want to take it from maybe a 9
out of 10 to a 10 out of 10, that's where 
something like this starts to make sense.
Used in a subtle way, it can bring out extra 
details and refine an image without changing
it too much. So it's not about replacing your 
workflow, it's about adding to it. And I think
that's where tools like this become interesting 
when you start using them with intention.
Well, all right, that's it for this one.
Let me know what you think, and 
I'll catch you in the next one.