I’ve taken bud bought from local dispensary, that has been tested, ran it through the hi grade tester and it’s always been Within 1%, they state that the accuracy is within + or - 3%, they don’t claim it’s perfect, it’s not, it’s a cool tool for the homegrower that isn’t rich.
1 Like
My read on this is it’s actually estimating the amount of resin coverage on the sample. If you used something that had a more mixed or balanced cannabinoid profile, my guess is it’s THC% would be way off.
A fancy version of what you could probably estimate by looking at it though a loupe after looking at a handful of known samples to have pints of reference. likely more consistent than a human estimate due to being less prone to bias.
1 Like
Exactly, it just uses AI to compare.
But then their just matching dispensary weed to known results for dispensary weed for certain strains on a library data base they must have. Like i said earlier different parts of the plant give different results so their is no way of 100% of knowing if the sample is even close to to the pic result system unless actually sent off. You can do the same by looking on line for the results of a strain an match it to what you have an say your accurate but its not factually accurate to the held sample unless tested.
1 Like
I wonder what would happen if you scanned a sample of black a5 or the mullum. Probably come up blank because no dispensary has them.
I think it uses trich field density and coloration, not strain by strain. You can put anything in there and will get a number.
2 Likes
I will check it out. 
1 Like
Keep us posted
Probably the best way to test the results of the Higrade Tester is to send a sample into a lab for analysis and compare.
1 Like
Can I test my sandwich?
It’s gluten free, I don’t know if that matters?
Will it test something like a fermented nug? Like the type of weed I got from Spain a couple yrs back that wasn’t dried to the core and was then packed together before sending, but retained that vibrant fresh green lime smell with some skunk undertone, and was sooo dank and really strong as fuck, but looked more like a sundried treefrog with little bud structure left? Would it accurately test that?
I’m just messing with ya, I couldn’t resist. 
1 Like
Sun dried tree frogs are actually what it was designed for 
1 Like
As I said I have been testing everything by the numbers coming out of the Dispensary and unless they are lying at the Dispensary then it works, I put this thread up to give people a heads up on if it works before buying it. If people want to call me a lier on the 3 months of testing it , then you have no faith in your fellow OG brothers and sisters or they are just trolling to rip others research and being negative. I never said it is accurate to lab standards! I said it was accurate to dispensary standards.
Believe me or not it’s your choice. Ignore it and keep paying hi dollar to get your bud tested 
3 Likes
It tests strains that are not in their data base, it tests it and if it doesn’t have the strain in the data base it sends it to them to add.
The GMO/ Cookies that I got at the Dispensery was not in Higrades system and it tested at 27, exactly what was on the package when I bought it. It sent the test and pics to Higrade for them to analyze and add to their system. And did same for several others I’ve tested that was not in Higrades system…
From what I’ve read it looks at color, trich density and a few other things. It says the test results are + or - 3 and I’ve tested several buds out of each purchase and the smaller buds always test out -2 THC less than the bigger buds on average.
I tested the live bud I had just to see what it said I didn’t say my live bud was an accurate representation, just that it did test out and that if the tester is accurate by what I’ve seen, that it should be good bud.
And believe me I’ve tested a shitload of different bud in the last 3 months to try and give my fellow OG brothers and sisters as impartial account of it’s accuracy to the bud bought at a dispensary.
Take it for what it is, it’s not like 70 bucks is outrageous.
3 Likes
For those asking how it works.
"
Our Science:
We are committed to science, and to make it accessible to as many people as possible. After 80 years of prohibition, in which scientific research on cannabis was restricted, there are still many gaps and misconceptions around the cannabis plant’s morphology and composition. Through patented technology and rigorous research, we are doing our best to assist anyone who would like to know this plant better with the goal of improving their lives.
Potency variation (heterogeneity) within cannabis flowers affects you every time you consume them
Cannabis flowers are like us – no two are the same. There are huge differences between cannabis strains, between plants of the same strain, between flowers on the same plant, and even inside each flower! However, today’s testing is done mostly through HPLC - a method designed for uniform substances (such as beverages), which is unfit to test cannabis flowers due to their inherent variability. The HPLC test is also destructive to the sample, which means that 100% of the flowers on the market were not directly tested! In fact, only 0.5% (average) of a batch is taken to regulatory demanded HPLC testing, with the primary goal of detecting pesticides and other chemicals – potency testing is secondary to that. And finally, even if you knew a batch’s precise average potency, it doesn’t mean you know the potency of the specific flowers you purchased, because of the huge variability within one batch. Sadly, it has become a “fact of life” - when it comes to cannabis flowers, people just don’t know how to consume them, and if they paid the right price for them.
Welcome to the world of trichomes
Some call them crystals, others glitter. Their scientific name, though, is Trichome (“hair” in Greek), and they are abundant throughout the plant kingdom. Stalked-capitate trichomes are the factory and reservoir of everything you love about cannabis. As the plant matures, they swell and change their color, while cannabinoids, terpenes and flavonoids accumulate and metabolize. Since they are on the surface of the flower, looking at them means looking at the plant’s chemistry. In fact, this is how growers know when to harvest cannabis, and there have been studies linking the appearance of trichomes in visible spectrum on the flower surface with its chemical composition. After realizing we can help growers know when to harvest, we realized we can help the consumers themselves to test their cannabis at home. Now, if you wanted to check sugar levels in green apples, for example, it would be impossible to do so just by looking at them – sugar is found mostly under the skin, where it is evenly spread. Measuring cannabinoids is different in that sense – the active compounds are right there on the surface, conveniently arranged in very concentrated “bubbles” – the trichome heads. Observing them in the right resolution (hence the HiGrade Scope’s magnification) is key to understanding their chemistry.
Training algorithms with chemical analyses
We are often approached by people who assume that in order to calculate potency, we look for the images in our database that are the most similar to the ones they uploaded. We don’t do that. Over the course of 3 years, we have collected many samples of cannabis flowers, collected images from each one, and sent each to an HPLC test in regulatorily approved labs in the US. This way we created a huge database of microscopic images of cannabis flowers and their corresponding chemical analyses. We then used this dataset to train a state of the art image recognition algorithm – deep neural networks (deep as in deep learning, a branch of AI). At every step of the learning process, samples are fed into the neural network – images and their corresponding potency (HPLC result). With each step, the algorithm tries to predict the chemical analysis which fits the image – and corrects itself if it was wrong. Over time, the network graph learns what features in the image matter and how much they matter. After millions of such steps, our algorithm found over 2000 such features, and reached a predictive ability of 85% accuracy. So what happens when you upload an image? Your image is being fed into our neural network, and the potency is calculated directly from it, depending on your image’s extracted features, and on each feature’s relative importance.
Let’s talk accuracy
It takes a whole machine to make sure the image we get from you is of high enough quality. We make sure it’s not blurry, not dark, that it’s taken with the right magnification, correct lighting, etc. Only a good enough sample enters the potency algorithm – the trained network discussed above. At the end of the training process, a test set of samples (which we have chemical analyses of, but the network was not exposed to during the training) was evaluated by the trained network. We calculated the deviation of the network’s prediction for each of the samples in the test set, and the resulting deviation average (RMSE) for the THC testing is 15% (statistical), which can be translated to +/-3% THC on average. As we mention on the app and this website, the accuracy value is true only when testing 3 different locations (to account for in-flower variability) of dry flowers originating in THC-only strains (normally having less than 1% CBD). It is important to note that like other testing methods, it has its limitations, and if you intentionally try to fool it – you may succeed. The true indicator for quality of any testing method is how well it performs when used according to its protocol, so we encourage you to put us to the test – as long as you follow the testing instructions carefully.
© 2022 HiGrade
HiGrade and their respective logos are trademarks or registered trademarks of MyCrops Technologies Ltd. in the United States and other countries. All other trademarks are the property of their respective owners. © 2020 MyCrops Technologies Ltd. All rights reserved.
"
HiGrade | THC Testing & Cannabis Growing Assistant | Our science.
4 Likes
As I said, take it for what it is. It may not be something for you, seems like AI tech is lost on some people. Personally I’ve tested it and found it worth the money
4 Likes
That may very well be the case, I didn’t read that into what was written, but if that is the case then home growers can still see how their bud matches up to dispensary’s best weed, so it is still a win, win.
3 Likes
You have a very expensive machine for taking pictures, sending them to MyCrops Technologies Ltd., and getting them to feed the pictures into an AI which has been shown a few million pictures of buds and cannabinoid datasets associated with them. It’s not useless, but I’d be very surprised if it can do more than a person can do by just eyeing the trichome density… as they’ve said, the real tests destroy the buds, so the real best way is to destroy part of your harvest, turn it into concentrate, and see how much concentrate you’re left with compared to the original weight. The downside to this method of testing is that you’re left with a bunch of pure-THC concentrate that you’re forced to dispose of, but there’s also an upside… you’re left with a bunch of pure-THC concentrate that you’re forced to dispose of.
3 Likes
I wonder if this would work on jungle sativas. I guess it would depend on if they are in the database.
1 Like
I’m thinking it would from my experience with the device. If it has trich it should work. It doesn’t work correctly on CBD based buds because it would mis-read it thinking it is THC based.
1 Like
I wouldn’t call it very expensive, it was 70 bucks , it seems cheaper than sending it in to get tested all the time. As for eyeing the trichomes by eye, I’d think that is less accurate and biased. When new tech comes out there is always going to be people that disbelieve and are set in their ways. I can look at it and say it’s good or bad bud but have no numbers to back it up.
With this I can atleast show that I have a device that has been designed to take some of the guess work out of it for customers.
2 Likes