LCD televisions are a technological miracle, but if they have an annoying side it’s that some of them are a bit lacklustre when it comes to displaying black. [Mousa] has a solution, involving a small LCD and a bit of lateral thinking.
These screens work by the LCD panel being placed in front of a bright backlight, and only letting light through at bright parts of the picture. Since LCD isn’t a perfect attenuator, some of the light can make its way through, resulting in those less than perfect blacks. More recent screens replace the bright white backlight with an array of LEDs that light up with the image, but the electronics to make that happen are not exactly trivial.
The solution? Find a small LCD panel and feed it from the same HDMI source as a big panel. Then place an array of LDRs on the front of the small LCD, driving an array of white LEDs through transistor drivers to make a new responsive backlight. We’re not sure we’d go to all this trouble, but it certainly looks quite cool as you can see below the break.
This may be the first responsive backlight we’ve brought you, but more than one Ambilight clone has graced these pages.
This type of lateral thinking is sadly under appreciated. Analogue control for the win!
Brilliant! Outside the box thinking at its best.
My thoughts as well!
That’s a lot of work to kill the Activate Windows watermark š¤£š¤£
But seriously it is a good hack.
https://iot-for-maker.blogspot.com/2021/08/led-12-displaying-hdmi-contents.html
f you had the HDMI signal why not skip all that rigamarole and just use the signal directly to drive the backlight rig? Lower cost, smaller size, lower power, less latency…
But this solution suggests buying a new projector just to get its LCD, then hack on a whack of photoresistors to measure its brightness poorly. Just…, why?
Seems like a “If your only tool is a…” kind of hack.
I assume it’s an easy way around HDCP. Also processing at the speed of an HDMI signal requires dedicated hardware. I assume if there was a good way to tap the lines going to the LCD panel itself that would work better.
Because then you need an electrical engineering degree, a powerful FPGA, and a lot of complex supporting hardware.
You’d need to process the HDMI data and average the brightness/gamma of all of the pixels in a given dimming zone, and employ some kind of smoothing algorithm that wouldn’t make it look like trash/give you hard edges when crossing into another zone. If you don’t devise a good (and very complicated) subsampling technique and brute force it, you need to process 497,664,000 pixels per second with a 4K TV at 60 FPS.
Or you could use the analog properties of light scattering and a photoresistor array.
The solution posited here is amazing in it’s simplicity. The complications involved in doing all that math and having all those dimming zones is why a lot of HDR LCD displays often look like shit.
You need an HDMI converter dongle to either USB or CSI, and OpenCV, and a Pi or similar. Engineering degree very much optional.
And then you get much bigger signal processing delay :-(
Those Chinese toy projectors are whole tens of $.
Yes, you could implement this using $20 Tang Nano 9k, minus the no flicker backlight part, but paradoxically it would take more work.
How much power is required to back light the “display” from the projector?
From all the reviews I’ve seen those local zone systems are awful due to the blooming unless they have a crazy amount of LED, and most commercials monitors have too few.
So in that sense I would not advise dong all this work.
However I do really like it as a standalone thing rather than as a backlight replacement, and that is a lot less hassle to do.
One note: We had chips with a dozen transistor drivers in a single package since what? The 70’s? So why do people keep using tons of individual transistors. Same for resistors actually, those also are found in single package with many same vale resistors.
Maybe we should start using ChatGPT to get part numbers for such parts. Ask it what chips have transistor/FET drivers in a package or what the part number is for resistor arrays.
I assume those AI things have that info right?
Extremely conflicted over what I feel about this project
On one hand its simple and does what it needs to.
On the other, its a rube goldberg machine type scenario and the engineer in me is cringing.
I would have thought enngineers liked rube goldberg machines.
Hmm..maybe the Hyperion project would worked simpler. https://hyperion-project.org/forum/ .instead of Ambilight clone (intended use) it could be used for local dimming with about the same BOM.
That’s the ‘easy’ bit of FALD. The hard bit is that you also need to modify the values of the actual pixels on the LCD (vs. the values of the incoming pixels) to compensate for the varying backlight brightness. Otherwise – as in the OP video – you get blown out highlights and bizarre misplaced falloff.
There was one brand of TV a few years ago that was using a low res B&W LCD display behind the main LCD to control the always-on backlighting. I think it was Samsung. That technique seems to have been abandoned now in favor of just using a lot of LEDs and controlling their brightness directly.
Practically speaking, this is something no sane person would attempt (though I might not be the best judge of sanity!). Nevertheless, it’s an absolutely awesome hackāeducational, informative, resourceful, and downright fun. It’s a testament to true maker spirit. Our boy really MacGyvered the heck out of this one!
The projector as the backlight? Color saturation?
Since the TV is cheap, maybe just use two. Stack the screens and use both backilghts.
Each “white” would only allow 50% of the light, but now you have twice the backlight, so that’s fine.
Each “black” pixel would now block orders of magnitude more light. No?