Roast Vision - a new and affordable roast level sensor for $299

Read about current and future offerings from the site sponsors.

#1: Post by espressovision »

Hi everyone!

My name is Zach and I'm the founder of Espresso Vision Inc., a small startup that develops hardware and software tools and programs for use in the coffee industry.

We just released our first physical product, Roast Vision, which was designed to be the most affordable roast level sensor in the market, for only $299! (not including taxes or shipping) It's also extremely lightweight and portable, never requires user-calibration, and only needs 1g of coffee per sample!

We believe that roast level sensors are an incredibly powerful tool for coffee enthusiasts, home roasters, and professionals that want to qualify the coffee they consume and roast, and hope that by providing a more affordable option, that this technology will become much more accessible than it has been in the past.

You can learn more about the product at our page here:

We're in constant development of new products, from espresso pressure gauge computer vision tools, to new air sensors for detecting first crack and the presence of TVOC's for ensuring a safe roasting environment.

I'll post updates here in this thread as we release new products and make updates, but please do not hesitate to ask any questions about anything! I'd love to explain our products and company goals.

User avatar
Team HB

#2: Post by TomC »

Welcome to HB!

This is interesting. Does it correlate to any other known scale for measuring roast color or is it entirely proprietary?

My next roast color analyzer I purchase will have to do both ground and whole bean coffee though.
Join us and support Artisan Roasting Software=

espressovision (original poster)

#3: Post by espressovision (original poster) »

Thanks for the welcome Tom!

Right now it only provides the Roast Vision proprietary output, but it would be very simple to map this to another proprietary scale, like the Agtron if I purchased a device that could produce such readings.

However, while it would be simple to collect this data, I'm not sure if Agtron has legally protected and/or copyrighted their scale, so the Roast Vision likely won't offer a conversion for a little while. Thanks for bringing this up as others have mentioned similar desires, so for the next iteration of the device I'll prioritize including a more well-known scale.

For whole bean and ground, do you use those measurements to determine how even the roast was throughout the bean? Or something else?

User avatar
Supporter ♡

#4: Post by hankua »

How does the sensor work?

espressovision (original poster)

#5: Post by espressovision (original poster) »

Generally, the device uses the principle of near infrared reflectance. By sending a known magnitude of light towards the ground coffee and measuring the magnitude of light that returns, one can determine the roast level based on the principle that lighter objects reflect more light than darker objects (lots of simplifications here but that's the general idea).

Specifically, this device uses a really cool open source photodetector from Sparkfun -

This component has multiple LEDs for illuminating the sample, as well as a sensitive set of photodetectors for detecting different frequencies of light. They're commonly used for pulse oximetry and small particle detection for things like smoke. This isn't programmed into the device for sale, but I was actually able to detect my pulse by placing my finger in the coffee sample cup, so it's been a really fun sensor to work with.

User avatar
Supporter ❤

#6: Post by Peppersass »

espressovision wrote:However, while it would be simple to collect this data, I'm not sure if Agtron has legally protected and/or copyrighted their scale, so the Roast Vision likely won't offer a conversion for a little while.
The Tonino can be loaded with a variety of conversion scales, including Agtron scales. I can't say that proves Agtron hasn't patented or copyrighted their scale, but at least it's a data point for you.

espressovision (original poster)

#7: Post by espressovision (original poster) »

It's certainly a good sign with the Tonino, and definitely a priority for us as we'd like the device to be able to be used to validate roast level for cupping, which at the current moment the SCA essentially requires an Agtron type output.

There was a lot of intention that went into designing the Roast Vision scale, that I'd like to add a few points here:

We started by designing a symmetric and balanced scale, where the dead-center medium is indeed at the dead-center of the scale.

On Roast Vision, this is between 17 and 18 (17.5 to be exact but Roast Vision only reports integers), with the darkest sample at 0 (-17.5), and the lightest at 35 (+17.5). Compare this to the Agtron scale where dead-center medium is 55, very light is 95 (+40), and very dark is 25 (-30).

While this may be inconsequential for more advanced users, we believe that having a balanced scale will be more intuitive for a general coffee consumer that may already be used to seeing other figures depicting roast level in a symmetric manner (5 empty circles for example from Light to Dark with the circle of the roast level filled in on a coffee bag).

We also wanted to ensure that the scale was meaningful in relation to the device's precision.

Take the Tonino for instance, which has a default range of 50-130, consisting of 80 unique units, and a precision of +/- 2 units. Alternatively, the Roast Vision, with a default range of 0-35, thus 35 unique units, has a precision of +/- 1 unit. In reality the precision is closer to +/- 0.65 units, but as the device only returns integer values, it would be inaccurate to report a precision less than 1 unit on the scale.

Instead of choosing to provide a scale with a similar number of units and precision as the Tonino (say 50-140 with 90 unique units and a precision of +/- 2 units), we chose to downscale the output range and offer very high precision instead. While there is great value in both having a large range AND high precision offered by many of the +$1000 devices in the market, the goal of Roast Vision was to provide a truly affordable option, and the sensor choice was one of the main compromising factors.

User avatar
Supporter ❤

#8: Post by Peppersass »

One big advantage I see over the Tonino is the lack of need to calibrate before taking measurements. I have the original Tonino in the tin can, and setting up the felt pads and calibrating, keeping them clean, storing them, etc., adds to the workflow. If I lost the pads or if they got dirty I might not be able to replace them (I don't think Tonino supplies them anymore.) The newer Tonino is better in terms of the setup and storage, and it might be a little easier to keep the color references clean, but the calibration process still takes a little time and the color references have to be kept clean.

I have mixed feelings about sample loading. You have an advantage in that the Tonino requires a full basket of tamped grounds, while you require just a small spoonful of untamped grounds. Your system fits in well with checking a sample immediately after roasting. However, the Tonino procedure makes more sense if you intend to pull an espresso after measurement, which is what I often do. However, it's not good if you intend to brew the coffee pourover, drip, immersion, etc. All in all, I suspect your setup makes more sense for people who roast and people who aren't mainly interested in espresso.

One improvement would be a rechargable battery, which would eliminate setting up the AC adapter and USB cable.

Need any beta testers? 8)

espressovision (original poster)

#9: Post by espressovision (original poster) »

Thanks! I chose to not include a battery to keep the device as low-risk as possible, both in terms of insurance premiums, as well as shipping requirements. However, I did design the device so that I could easily include a battery in future models, and have a design already with the current housing that would provide 5-7 days of battery life, so it's a very energy efficient system. If you're into Arduino at all, there's a list of parts at the bottom of my product page with some links to the Open Source hardware I'm using from SparkFun.

Regarding sample loading, I tested a few different designs, including a similar model that would measure the sample in the same way as the Tonino since it is such a great design, but due to the fact that I am 3D printing the enclosure, which is NOT food safe, I decided to make it clear that the coffee sample for testing is not to be used for brewing/consumption.

Often if I'm brewing a pour over, or even espresso, I'll grind the coffee for my brew, and then turn my grinder to the finest setting, purge a few beans, and then grind a tiny sample at the finest setting to measure in the Roast Vision.

No beta-testing needed, but I'll DM you a coupon code for $50 off as a thank you for all of the feedback - I really appreciate the interest and hope that this first device and the future iterations will be useful to those in the community.

User avatar
Supporter ❤

#10: Post by Peppersass »

I'm major into Arduino -- used one to add flow and pressure profiling to my GS/3 :D.

Now that you mention setting the grinder very fine for measurement, that's actually another advantage over the Tonino method. Some have commented that you have to grind very, very fine to get accurate and consistent results with the Tonino. That means you may not be able to use the grounds in the basket after measuring. Personally, I've always measured at the grind setting I'm using to pull the espresso, and haven't taken the time to compare readings at different grind settings.

So if a really fine grind is essential for consistency and accuracy, your small sample requirement is a definite advantage.

I'm game. Order placed.