|
Gamescom 2026: Scentbar aims to bring scent to games. After years of attempts, this is the first time I feel it could work9/6/2026
Who wouldn’t want to smell lilac and gooseberries when replaying The Witcher?
|
Games have long since stopped trying to engage players through visuals alone. Technology has gradually improved sound, surround sound, controller vibrations, haptic feedback and various other ways of immersing players in virtual worlds. One area, however, remains practically unused: smell. There have already been several attempts. In the past, I had the chance to try various technologies that attempted to bring scents to games or virtual experiences. Most of them were interesting ideas that very quickly raised the question of whether they had a practical place in the home and whether they could work accurately enough.
Scentbar is another attempt. This time, however, after the presentation and trying it for myself, I feel for the first time that similar technology could actually work. Not because it is already perfect. Quite the opposite: during the presentation, we also saw moments when the system simply got things wrong. The basic idea, however, seems considerably more practical than many previous attempts. Scentbar does not want to wait for developers to implement scents directly into their games. It is trying to recognise them on its own.
The name Scentbar itself gives a fairly good idea of what the device resembles. In terms of design and size, it is something like a smaller soundbar. Instead of speakers, however, it contains slots with individual scents. The device can hold ten different scents at once, then selecting the appropriate one depending on the situation. The company currently has approximately 33 scents prepared, with more to be added over time. The scents are divided into different packs. During the presentation, we saw, for example, the Nature Pack, Racing Pack and Action Pack.
The Racing Pack contains scents such as rubber, fuel, engine oil, asphalt and the smell of a new car. The Nature Pack naturally focuses on nature, forests, the sea and other environments. The Action Pack, meanwhile, works with effects that can evoke shooting or explosions, for example. During gunfire, one scent was activated that some people compare to fireworks.
The basic idea is that you can choose which scents you want loaded into the device. If you mainly play racing games, you can prepare Scentbar specifically for them. If you want to use the device with fantasy games, you can create a different combination. The creators also want to gradually prepare preset packs for specific game worlds.
During the presentation, they showed a concept setup for The Witcher, in which the combination would consist mainly of natural scents, several action elements and other specific associations. And yes, naturally I had to ask whether Yennefer’s scent would be included—in other words, her famous combination of lilac and gooseberries. It is not yet a specific official scent created for the game, but the creators suggested that they could create a similar association among their scents.
The most interesting part of the entire system, however, is in the software. Scentbar does not require a particular game to have official support for the device. The entire system works through local AI-based image recognition. In the software, you simply select a specific window or the entire monitor, and the system starts watching what is happening on the screen.
The AI then recognises objects. It works with shapes and colours. With prominent text, it can also use textual information. If the system recognises a tree, for example, it can activate the corresponding natural scent. If you are by the sea, it can activate a scent associated with a marine environment. During the presentation, we tried the system directly in Minecraft, where the AI watched the screen and gradually activated different scents based on what appeared on it.
And this is where the technology’s greatest strength, as well as its current limitation, became apparent. When the AI correctly recognises an object, the result is surprisingly good. When it gets it wrong, you get a completely different association. At one point, we were looking at a red tree. The AI, however, interpreted it as lava. Instead of the expected fruity or natural scent, Scentbar began releasing the smell of burning. It may be an amusing example, but it also shows very clearly where the device’s future will be decided. Not in whether Scentbar can release a scent. That already works. The decisive factor will be how accurately the AI can understand what is currently on the screen.
The creators are working with a threshold system. The AI does not have to react immediately to every object that vaguely reminds it of something. In the software, you can set the sensitivity and confidence level the system must reach before activating a scent. At a lower threshold, it will react more often. At a higher one, it will be more cautious and activate a scent only when it is more certain. The creators say the system currently works correctly in approximately 95 percent of cases. The figure should of course be taken as their own estimate from the presentation, but the demonstration itself suggested that recognition is already fairly impressive with ordinary, unambiguous objects.
The biggest problems may occur with stylised graphics. Minecraft is a good example, because very simple shapes and strong colours can be interpreted in different ways. A red object does not have to be lava, even if it may appear that way at first glance. The creators are therefore preparing additional filtering options. For example, the system could learn to ignore certain specific objects or combine the recognition of several elements to avoid similar errors. And in my opinion, this will be the most important part of further development. The hardware may be ready. The scents may be accurate. The fans may work. But if the AI cannot understand what it is seeing well enough, the result will inevitably fluctuate between highly impressive and completely absurd.
Scentbar does not have to work with just one scent at a time. If the AI recognises multiple relevant elements, the device can activate several scents and create more complex combinations. During the presentation, the creators gave the simple example of pineapple and coconut. If the system recognises both objects and both reach the required confidence level, you can get a combination reminiscent of a piña colada.
The same can work in a game environment. Imagine a campfire in a forest. The system can combine the scent of the forest and smoke. If a burning robot appears on the screen, a metallic or mechanical scent can be added alongside the smoke. Scentbar is therefore not yet a system like an RGB printer that could create any scent by mixing several basic components. Each scent is separate. The device can, however, combine them.
And the ability to mix scents is one of the areas the creators want to develop further. Their long-term vision is a system that can create increasingly precise and subtle combinations based on what is happening on the screen.
One of the things I was most interested in during the presentation was the precision of the scent release itself. With similar devices, the problem is straightforward. If you release a scent into the entire room, it stays there. The game may meanwhile move from a forest to a city, from the city to a garage and then to a burning house, but you may still be able to smell the previous environment. Scentbar attempts to solve this problem with directed airflow.
Behind each cartridge is a small fan that activates when the AI decides that the particular scent needs to be released. The air then passes through the specific cartridge and is directed forward towards the player. The creators say that the airflow system is one of their patented solutions. The important thing is that the oils or scents are not actively dispersed throughout the entire room. They therefore should not remain in the air as strongly. The scent is directed relatively straight towards the player’s nose and disappears fairly quickly once activation ends. During testing, the release of scents was surprisingly precise and fast.
Of course, this does not mean that you receive the scent with the same precision with which you see a pixel on a monitor. Smell works differently, and a lot also depends on when you happen to inhale. The creators are working with this natural human characteristic as well. The AI can recognise an object and start the fan in approximately half a second. The scent itself then has to travel a short distance to the player. People naturally pause between breaths, however. The resulting sensation therefore does not feel as delayed as the need to first analyse the image, then activate the device and finally deliver the scent to the player’s nose might suggest. In practice, Scentbar’s response was one of the more pleasant parts of the entire presentation.
In my opinion, Scentbar is primarily a PC peripheral. Not because it cannot work with consoles. It can, and very easily. But physics cannot be completely fooled. With a computer, you usually sit relatively close to the monitor. Scentbar can be placed directly in front of you, and the scent has only a short distance to travel to your nose. The situation is completely different with a television in the living room. If you have Scentbar next to the TV and sit two or three metres away, the intensity of the experience naturally will not be the same.
In theory, you can increase the fan power and scent intensity, but that gradually brings you back to exactly the problems the device is trying to avoid—dispersing scents throughout the space and making them remain in the room for longer. The creators are aware of this. For consoles, they are preparing an HDMI synchronisation device that would allow the console’s image to be analysed locally without requiring direct game support.
The system itself can therefore work with practically any image source. With consoles, however, the question of placement remains. In the future, the creators are also considering solutions similar to surround systems that could bring the scent source closer to the player. In its current form, however, Scentbar makes the most sense to me on a desk next to a PC.
The basic Scentbar is set to contain ten cartridges, and the device’s current price starts at €299. The basic package therefore includes ten different scents, which you can choose according to your preferences and the games you play. The full set of cartridges should last approximately 60 hours of use. Actual endurance will of course depend on intensity. If you set Scentbar to trigger scents frequently and run the fans more intensely, the cartridges will be used up faster. The creators say their goal is for a standard set to last approximately two to three months with regular gaming.
Replacement packs are expected to cost approximately €39, which at the projected endurance would mean roughly €10 per month. In addition to simple sensitivity settings, the software also offers more advanced options. You can adjust, for example, the length of time the fans run, the interval between activations, intensity, fan speed and even the device’s LED lighting. The device has its own backlight, which already uses different colours to indicate the scents being released.
On the other hand, the creators also want to prepare a simple plug-and-play mode in which the user simply launches a game and selects the appropriate profile. Ideally, you would not have to manually set what the system should recognise before every game.
The system also has interesting potential for streaming. Since the AI does not have to analyse a specific game but can watch the image itself, it could theoretically work when watching a stream as well. Imagine that a streamer is playing a particular game and has created a scent profile. The viewer can select the same profile, and Scentbar will then analyse the image appearing on their own monitor. This is not a system in which the streamer would have to send separate information about which scent should be activated. Each Scentbar analyses the image locally.
In my opinion, this is one of the reasons why AI recognition potentially has greater significance than traditional integration directly into games. Developers do not have to do anything. The game does not need an update, and the device can attempt to work with content that was never designed with scents in mind.
This was the point at which the presentation began to shift, for me, away from the question of whether players would want to smell forests, petrol or smoke. Scentbar could be an interesting peripheral for players. But perhaps there are people for whom it could open up entirely new possibilities. During the presentation, one of the creators simply opened a web browser and searched for an image of a pineapple. A pineapple appeared on the screen, the AI recognised it and the smell of pineapple began spreading from Scentbar almost immediately. It was a simple moment, a quick demonstration of the possibilities.
But that was when I started thinking that players may not be the most interesting group that could benefit from this kind of technology. What about blind people? The internet already offers many ways to make content accessible to people with visual impairments. Websites can have text descriptions of images, which screen readers can then read aloud. An image can be described through sound. But one more sense is still missing. Imagine if a website with a photograph of a forest provided not only the information that the image contained a forest. The computer could also add its scent. A photograph of the sea could produce a marine scent. Food could produce its characteristic smell.
With a photograph of a pineapple, Scentbar would simply do what it demonstrated during the presentation. This does not mean that smell can replace sight. Of course it cannot. It could, however, provide another layer of information and association. And in some cases, that layer could be extremely powerful. The creators have already tested similar uses and acknowledge that the technology’s possibilities in this area are very broad. The foundation will once again be recognition accuracy, and especially gathering data on how people use the device and which scents are useful to them.
For me, however, this moment significantly changed how I viewed the entire product. Suddenly, I no longer saw it merely as another attempt to add a new effect to games. It could be a technology capable of supplementing images with a new type of information.
The creators are well aware that Scentbar is not entering a completely empty market. There have already been several similar attempts. A large proportion of them failed to find a sufficiently convincing reason why an ordinary person should buy and use such a device over the long term. The technology itself is not enough. The right market needs to be found.
The creators are therefore currently focusing mainly on simulators. That is where they see groups of players who could be most interested in scents. Farming Simulator could benefit from the smell of nature, soil or various environments. Just imagine the whiff of manure when driving past it on a tractor. A racing simulator, meanwhile, could benefit from rubber, fuel, oil or asphalt. In games that try to simulate a specific activity, adding another sense may feel more natural than it would in an ordinary game. And perhaps this is precisely the right direction. Scentbar does not have to convince every player. It does not have to become a new standard that everyone keeps next to their monitor. It is enough if it finds a group of people to whom it can offer something that no other device can.
After the presentation, I do not feel that Scentbar is a finished, problem-free product. Quite the opposite. The current prototype still clearly shows its origins. The construction is visibly made using 3D printing, and the device does not look like fully finished consumer hardware. That did not bother me too much during the presentation, however. What mattered more was that the basic technology worked. The scents activated relatively accurately. They reacted quickly. And thanks to the directed airflow, they did not linger in the space as long as I expected. Moreover, the construction is really simple, the scents are easy to replace and they are not complicated cartridges. They are essentially just small cartridges with a circular opening, containing scent-infused beads made by European perfumers. A small fan blows air through them, directing it through the centre of Scentbar towards your nose, and the effect is ready. It really is that simple.
The biggest question remains the AI. When it correctly recognises a scene, the result is impressive. When it interprets a red tree as lava, you get the smell of burning instead of a fruity scent. And it is precisely between these two moments that Scentbar’s future will be decided. After years of trying various technologies that sought to expand the gaming experience with smell, this is the first time I feel there is a real chance.
Not because Scentbar can already perfectly convey the scent of every game world. But because there is finally a practical system that does not have to wait for developer support. It only needs the image. If the AI gradually learns to recognise objects, environments and their combinations more accurately, Scentbar could work practically anywhere. In games. On streams. In videos. On the internet.
And perhaps it will ultimately find an even greater use outside gaming than directly among players. Personally, Scentbar is therefore not just a question of whether I want to smell lilac and gooseberries while playing The Witcher or asphalt while racing. It is one of those products where I am more interested in where it could be in a few years.
For now, it is a device starting at €299, with ten scent slots and approximately three dozen available cartridges. The technology still needs to mature, especially in the area of AI recognition and the final hardware itself. But in my opinion, the foundation is right this time. And if the creators manage to refine the accuracy of image recognition, Scentbar could be one of the first devices of its kind that does not function merely as an interesting demo at a technology exhibition. Perhaps then we could finally start talking about smell as another genuine input into digital worlds.
