Nvidia Image Sharpening : Budget Gamers Guide to Squeezing Higher Frame Rates with Minimal Visual Compromise (Guide/Analysis)
First of all , background. So being the low-end gamer that I am , I’m all for new toys both AMD and Nvidia are throwing at us that could help improve performance on our current GPUs. Quite recently , Nvidia turned a lot of heads when announcing one of their new features called “DLSS” which claims to be able to upscale lower resolution images to higher resolution ones with minimal loss in detail thanks to their “deep learning technologies”. The main thing they were focusing on was upscaling 1440p images to 4k , and being the low-end budget oriented gamer that I am , it wasn’t going to be anything useful for me. Fast forward a couple of months , AMD announced “Radeon Image Sharpening” which pretty much made the same claims as Nvidia’s DLSS , that is to improve visual quality at lower resolutions to make them look closer to higher resolution ones. You can see what their claims were in the following video.
Soon enough , Nvidia followed suit in releasing their own version of Image Sharpening , probably thanks to AMD making their version open source. And in typical low-end fashion , I wanted to find out how it would perform in sub-1080p resolutions. If it claims to be able to make 1080p quality closer to 1440p , can it make 900p quality anything close to 1080p? We shall all find out.
What is the goal ? The goal is to cut down on resolution a bit without actually sacrificing on Image Quality by using Nvidia Image Sharpening to retain close to native resolution image quality.
Does it work for every game ? Sadly no , it can only be used for good effect in games that allows Temporal Anti-Aliasing (TAA/TXAA). While Image Sharpening can be applied to games that doesn’t have TAA , from the games that I’ve tested , it seems that it won’t be able to retain the details lost and instead it will make the game look much worse. But worry not , TAA is becoming an increasingly popular choice of anti-aliasing because modern up-scaling technologies such as DLSS also requires Temporal Anti-Aliasing to be present.
Define “Minimal Compromise” ? Very minimal , the up-scaled image is never going to look exactly like the native resolution image. But it comes very close and looks a lot better than playing at a lower resolution without any sharpening. Image Comparisons will be included down below.
How much of a FPS gain can I expect ? Depends on the game , but you can generally expect around 20–30% increase.
Now let’s get into some comparisons !
First let’s take a look at one of the more demanding games. Red Dead Redemption 2. Here we are comparing 1080p quality against 75% Resolution Scale of 1080p (roughly averages to 900p).
In the video above, I am alternating between turning the Image Sharpening ON and OFF. The timestamps for each switches are as follows.
(0.00 — ON), (0.09 — OFF), (0.23 — ON), (0.29 — OFF), (0.36 — ON), (0.41 — OFF), (0.45 — ON)
Here one of the things that immediately stands out is the quality of the textures on John’s sleeve. In the 75% resolution scale it looks extremely blurry but adding sharpening to it have managed to get it looking as almost as good as the original 1080p quality. Then when looking at the house we can see this effect clearly , sharpened image have managed to recreate the sharpness and details in 1080p image as closely as possible. Let’s take a closer look at the mountains in the image.
As you can see the mountain looks much much better in the sharpened image and in the non-filtered image , all the trees just blend in together while you can distinctly separate them in the sharpened one.
I’m sure the question on everyone’s mind is right now is , what are the performance gains like ? Let’s take a look
As you can see , you can gain a decent performance gain upwards of 20% without having to sacrifice too much on the quality.
Next up , Assassin’s Creed : Origins. Here we’ll be comparing 1080p against 70% Resolution Scale of 1080p with the sharpening filter on and sharpening filter off.
Again here we can see that with the use of Image Sharpening we have managed to regain some of the details that were lost when turning down the resolution scale. This can be clearly identified when looking at the textures on the walls and specially the roof. Even the leaves in the trees looks extremely more detailed in the sharpened image.
Now let’s dive into some extreme cases , dive down to so much as 720p. (44% Resolution Scale of 1080p). For this example we’ll be looking at Shadow of War. I can’t stress this enough , cutting down to 44% Resolution Scale , THIS IS NOT SOMETHING I RECOMMEND ! It’s not going to look anywhere near as good as the 1080p image. Only do this if you’re extremely desperate for more frames. And we are going to try our best to make the 720p game-play look as pleasant as possible. First let’s take a look at a best case scenario.
You can immediately see a huge difference between the 720p image and the 720p sharpened image. The sharpened image might be a little over-sharpened for some people’s liking but you can tweak it to your liking. When looking at the textures on the building as well as textures on Talion’s cape , you can see the blurriness is gone and and the tiles on the building looks more distinctive. Now let’s take a closer look at those building textures.
One of the undesired side-effects here is the snow as you can see they look a little more highlighted than they should be. But on the positive notes , the quality of both the building textures and the orc are day and night between the two images and things are way less blurry and detailed in the sharpened one.
This all looks fine and dandy , so why didn’t I recommend cutting the resolution all the way down to 720p? The reason is noise. As I mentioned above , the image was captured at an ideal situation to showcase the strengths of Image Sharpening. To get the 720p image to as close to as sharp as the 1080p , I had to put the Sharpening filter strength at maximum. In doing so , it adds a lot of visual noise to the game. I’ll show an example.
You can see on the left side of the screen , entirety of all the textures are filled with undesired noise. This makes the game looks horrendous. You might not have any blurry textures , but the sharp textures you’re going to get are going to look like they are suffering from some serious skin condition.
GUIDE
First thing you’ll have to do is to make sure you have your Nvidia In-Game Overlay enabled. To do this , open your Nvidia Experience Application , click on settings (the little cog button on the top-right) and make sure your in-game overlay is switched to on.
Now that you’re set , open up the game you want to apply Nvidia Image Sharpening to. Navigate to the settings menu and set the anti-aliasing to Temporal Anti-Aliasing , which is also known as TAA or TXAA.
And when you’re done , the next step is to turn down the resolution a notch. If the game you’re turning down the resolution has a Resolution Scaling , this makes our job much easier. I recommend setting the resolution scale to 70%-75% . You can go higher if you want but make sure not to go below 70% because then we’ll possibly notice a lot of visual degradation even with sharpening on. If your game doesn’t have an in-game scaling option , you’ll have to adjust the resolution directly. When doing this , first find out the aspect ratio of your native resolution and only change the resolution to a resolution with the same aspect ratio. For example if your native resolution is 1920×1080 , it means your aspect ratio is 16:9 so you can change your resolution to 1280×720, 1366×768 or 1600×900. I highly recommend changing it to 1600×900 , going any lower can have significant visual degradation and loss in detail that Image Sharpening won’t be able to fix.
Now all that’s left to do is to tweak the image sharpening to your liking. Head on in game and press ALT + F3 and it will bring up Game Filters screen. Here click on 1 and then click on “Add Filter” and you’ll see a list of filters that you can add. Scroll down to Sharpen and select it. Now you can adjust the sharpening to your liking. I recommend staying between 50% and 70% , over-sharpening can add a lot of visual noise to the game and having the sharpening too low can make the game look less detailed and blurry because we are playing at a lower resolution.
(Optional Step : to make sure your settings survive driver updates)
When you figure out the sweet-spot , open the Nvidia Control Panel. Then in the “Manage 3D settings sub-menu” , click on the Program Settings tab. Then select the game you wish to apply the filter to from the drop-down menu under “Select a program to customize”. If the game you want to apply the filter doesn’t appear on the , you can add it manually by clicking the “Add” button next to it and select the the program from there. If it doesn’t appear there either , then click on the “Browse” button and navigate to the executable file of your game (normally located inside the bin folder inside the game folder). And when you’re done with that step, Click on Image Sharpening from the options below and adjust it to the values that you found the best when you were testing it out from the in-game overlay. Then click OK and then Apply.
Congratulations ! Now you’re utilizing Nvidia Image Sharpening.
Final Thoughts and where can we go from here ?
I’m fairy impressed with this tool , but there’s also room for improvement. One of the main issues that were present is the visual noise , specially in higher strengths of sharpening. But other than that , it’s a pretty sweet deal. One has to wonder how this can be implemented and used along with other technologies, specially Dynamic Resolution. A feature that a lot of games seems to adapt these days. Dynamic Resolution works by lowering the in-game resolution in graphically intensive situations to keep the frame rate stable. Judging by how fast the Sharpening Filter can be turned on and off , I think there’s a lot of potential for this to work along side of Dynamic Resolution to give us a great quality to performance value if it gets implemented properly.
If you made it into this tutorial/guide/analysis so far , I thank you for your time! This is my first time writing a piece such as this , so be kind enough to leave some feedback if you have the time. Have a great day !
