Infiray XH09 X2 series, discussion

[nikitasius]

Infiray X2

• Resolution: 256x192@12µm
• Refresh rate: 50Hz
• App: Thermal Eye X (gplay)

Got mine on Aliexpress in official store:

• store URL https://infiray.aliexpress.com/store/911200160
• X2 TIC https://s.click.aliexpress.com/e/_EGW9qXp

My tests: https://drive.google.com/drive/folders/12hrdmE1RqeafaK5wba5T0-0Vi8uT--Kz
Compared to covid Hikvision https://www.eevblog.com/forum/thermal-imaging/infiray-xh09-x2-series-discussion/msg5098434/#msg5098434

Youtube:

• Pretty unstable guy, don't buy from him
• Some random review

Image is pretty good (compared w/ Seek images i've seen on the forum or over the internet). I use it for bird hunting (no gun, just binos & my DSLR) and for PCB inscpections. Camera very sensitive i really like it. But the app is poor, full auto mode, you can change just the materials. Hope it will be changed in future (setup the temperature ranges etc).
Big issue: temperature values, mine display +7..+10°C more than should. Idk why, maybe due app or firmware. But i don't use it for measurements, just for the image. Also i can't fix it as is into handle, due they sent me handle w/ small soft pads.

Reviews on EEVblog:

• 2023-10-14@40days, nikitasius : post#35
• 2023-12-07@blargg: post#45

[nikitasius]

Looks like real name is XH09 and X2 just a sub-label due default x2 zoom.

Also folks say default zoom is x2 digital, some say optical.

Also this morning i just connected it to VLC under my debian and got this. I haven't good picture (just stop frames, but then i restarted VLC i got just GREY image). But whats interresting:
There are 25fps and 256x384 px. Is it have 384x256 bolometer? Or maybe firmware shots 256x192 on 50hz, sticks to chunks 256x192+256x192 = 256x384 and sending it at 25Hz rate?
From aliexpress: folks had issue, temp mode on old app had 9Hz, after being fixed temp mode display now 25Hz. Model have x2 name. So could it be 384x256 which is partially used to increase Hz?
Some flir cameras increase Hz when you decrease the image height so could it be 256x384 (this orientation) cut in a half?.. too many questions but no teardown yet.



UPD: 2023-09-05

> These cameras return a single image where the top 256x192 pixels represent a dynamically balanced grey scale image, and the bottom 256x192 represent the raw temperature data.

About P2 Pro https://www.eevblog.com/forum/thermal-imaging/infiray-p2p2-protopdon-temperature-data/msg5039713/#msg5039713

[nikitasius]

Could be captured via virtualdub under windows and gives that.

[Fraser]

Lens is ~19mm giving an approximate X2 magnification over the standard T2 models, hence X2 nick name.

The camera software offers up to x15 digital zoom, hence confusion amongst some regarding optical of digital magnification.

The measurements accuracy should be as stated in the specifications. Any serious deviation from those specifications suggests a fault in the system, be it hardware or software.

[nikitasius]

About measurements.
They could solve it "quickly"

[nikitasius]

Well, thats weird

[nikitasius]

Here are the measurements.
Chair w/ pip - mine
Glove & ice - from a random guy from discord.

Also, seen that Fraser in M600 topic told:
> if a manufacturer skimps on the ambient temperature measurement system or calibration of the camera, errors will creep into the measurements.

And Infiray asks for ambient temp.. i pretty sure it screw the measurements

[nikitasius]

Sufraces which are >100°C
Video is in the ZIP file (can't upload mp4 here)

Weird effect 🤔

[Fraser]

Have a look at my review of the Infiray P2 Pro. In that review you will see the results of the camera viewing a BlackBody that exceeds the recommended maximum temperature of the selected sensitivity range. Once I switched the P2 to its lower sensitivity (higher temperature) range all was well again.

https://www.eevblog.com/forum/thermal-imaging/review-infiray-p2-pro-thermal-camera-dongle-for-android-mobile-phones/msg4575070/#msg4575070

Fraser

[nikitasius]

Hard to get answers from infiray..

[nikitasius]

Well accuracy is shit poor.

On a skin collage it use wrong preset (default), skin screenshot is added.

I contacted them.

[nikitasius]

Read topics about Seek & Infiray and did a cold test (big cold plastic from -19°C freezer).

Video from the app (bit compressed) https://drive.google.com/file/d/1BbTklsxRQ0HlClVSbEeY8De51A0pLu4R/view?usp=drivesdk

Screen record https://drive.google.com/file/d/1BFpC4o9UiIAmaa8D-fIpqaAmHx28bM-l/view?usp=drivesdk

If there are some anomalies? 🤔
I see some "static" traces, but they are here even after clicking shutter button.

UPD: one more, uncompressed
https://drive.google.com/file/d/1BnPfXpxvf7UkNpzKw10x9HKKue53Tnge/view?usp=drivesdk

[nikitasius]

Same weird effect on that vidéo also



You can see it on the teapot. I asked them wtf is that. Interresting to know.

P.S. youtube video in zip file also.

[ArsenioDev]

Yeah that's ADC saturation on the ROIC. Essentially the temp range is too high for the conversion gain set for the microbolometer TIA, thus being too high to register past maxout.

[Bill W]

Yeah that's ADC saturation on the ROIC. Essentially the temp range is too high for the conversion gain set for the microbolometer TIA, thus being too high to register past maxout.

I would have expected ADC (or number space) saturation to show columns static to the image frame, ie the column with the most gain goes bad first. 
Most of that seems to be following the kettle about suggesting it might be real ?
(Only looked at the YouTube)

[nikitasius]

ArsenioDev & Bill W

There are the video i made right now:
https://drive.google.com/file/d/1DMcm_OkvWY9lLH5BO7fjd1LUCqN7tZJt/view?usp=drivesdk

I connected TIC, wait for 5-7 min, calibrated on a cold iron surface (for ironing) and turned on an iron.

Second 23 i click on calibration button (shutter) and get pixels
Second 28 i hide with my hand (and move it few times to see pixel effect).

My good friend say it looks like postprocessing issues.
In that case it could be fixed (if i need to film very hot surfaces) via IR filter (external to TIC).

[ArsenioDev]

Yeah that's ADC saturation on the ROIC. Essentially the temp range is too high for the conversion gain set for the microbolometer TIA, thus being too high to register past maxout.

I would have expected ADC (or number space) saturation to show columns static to the image frame, ie the column with the most gain goes bad first. 
Most of that seems to be following the kettle about suggesting it might be real ?
(Only looked at the YouTube)

Column washout isn't as prevalent these days, modern ROICs reduce bloom and that noise drastically.
If it was "real" texture, aspect angle change would have it change like laser speckle, fixed pattern suggests inherent to detector.

[nikitasius]

Screencapture:
Image collect noise or idk what, but after shutter goes better.
Whats weird: some dot patterns are pretty visible.

https://drive.google.com/file/d/1DazK1n6hZ3CQ-1-k_NMPLdKmFT_no9O4/view?usp=drivesdk

[nikitasius]

Frozen it a bit w/ a gas. Like folks says, it becomes noisy.

[nikitasius]

Well, left video is from today.
Right video is from 1st day (almost 3w ago).

https://drive.google.com/file/d/1GgAh4UX8kjmPMHAk0bxAr20phfeKYgWR/view?usp=drivesdk

On video you don't see a lot, but you see well on screen captures i shared on gdrive & here.

What i personally think:
• image degraded, i see static noise patterns, like TIC going to have HW issues w/ the bolometer
• from the box software is pretty poor (android)
• website is pretty poor in case of specs and that sucks

Will i buy another Infiray? Well, NO.
Is the price is justified? Finally it cost for me 305€ (got 38€ back due wrong holder).
305€ for 256x192? Well, NO. I prefer to ebay other used brands but not the infiray from that price category (and guess anotherones).

Infiray sucks. If you can get flir/fluke/hikvision/agm for same res for same price (handheld models), take them.

Price of that TIC must be 200 bucks.

[Fraser]

Your posts regarding the XH09 make interesting reading. Thank you.

The XH09 is a little different to the other Infiray camera dongles. It claims 50fps and uses new image processing algorithms to enhance the image. I do wonder whether this particular camera is suffering from immature development as other Infiray camera models seem to perform well. It remains a mystery why your camera would appear to have suffered an increase of noise content in the produced images over time. If a microbolometer loses its vacuum it suffers a massive decrease in thermal sensitivity so I doubt that is the issue here.

I did write to Infiray some time ago, and again recently, offering to test and review any of their cameras for this forum but sadly received no response from their sales team. It would have been good to carry out technical testing on their product range to establish performance and differences in performance between models. Maybe they do not want such technical testing and prefer the usual YouTube user review format ?

When I studied the XH09 I saw a dongle that was trying to impress the marketplace with double the usual frame rate and claims of custom video processing. Sadly it may be a flawed product as we know that IRAY Infiray microbolometers are capable of decent imaging. There is also the possibility that your unit was faulty from new. The temperature measurements were very inaccurate and that can be a sign that something is wrong with the camera hardware or its calibration.

I am sorry that this buying experience has not been good for you.

Fraser

[nikitasius]

Hi Fraser, you don't need to be sorry. I bought it myself w/o being affected by any review.
Also there are at least 1 more person with température mesurement issues who own xh09.
Sure folks who do casual reviews will never tell about the issues they could have.
Also idk if there are real 50Hz. I will build a rotating demo, capture w/ laptop and compare 4 frames together.
Cause if it use mini256 (as xh15 in theory use mini384) it suppose to have 25Hz.

[nikitasius]

1st connection,
* camera init guess: Screenshot_2023-09-17_13-11-45.png
* camera in working mode: Screenshot_2023-09-17_13-11-59.png

Screenshot_2023-09-17_13-17-38.png - the pattern which appears also when i point at hot objects. This pattern appeared when i changed 256x384 to 256x192 resolution.
Video is in this file Screencast 2023-09-17 13:16:08.mp4.zip  (or gdrive https://drive.google.com/file/d/1rIXqot9rt6UCO5JpgukNqVYSEulotWsE/view?usp=drive_link )

A flawed bolometer? Bad soldered (so they are killed part of pixels)?

[nikitasius]

That happen when i change the FPS from 25Hz to 50Hz

video in zip or on gdrive https://drive.google.com/file/d/1gdVfV4ovKGwyk2ZKPfFu6raUyY-BBXej/view?usp=drive_link

[nikitasius]

Rolled into a film and put inside my frigo (+3°C) for a while.
Removed the film, connected w/ my pixel7 and point at the wall. Became noisy while was warming. Different pixels had different noise level, but, looks like there are no "dead" pixels which will be cold or hot. Just less hot guess.
Got some condensation on a lense but it doesn't affect the test.

Here is a screencast (gdrive): https://drive.google.com/file/d/1HDz2m1trhZNHNKaMxU8DigSPLcrOci_3/view