Random Thoughts

Posted on

Up-scaling images.

I keep getting ads for Topaz up-scaling software, and have wondered if it is any good. Not that it matters, since I am running on Linux computers, and Topaz is Windows. But recently I found an open source up-scaler program names Upscayl, which runs on Linux, Windows, and Mac. So I installed it to do some testing. First thing I discovered is that my built in graphics card wasn’t powerful enough to run the software. So I had to buy a Nvidia GPU. I bought the least expensive one I could find, and it runs the software just fine.

My simple question is how does this software compare to scaling in Gimp. For example, if I take an image and scale it 400% in Gimp, how does that compare to scaling the image 400% in Upscayl.

So I took an image I have of a hummingbird.

I will upscale it and then crop it to the same dimensions, 1920×1080 pixels.

Here is the new image up-scaled by Gimp.

And here is the image up-scaled by Upscayl.

Upscayl did create a sharper image. Interestingly, the image it created is at a lower resolution, 300×300 ppi, where Gimp left the image at 350×350 ppi. But the Upscayl image is slightly larger, 1.3MB compared to 1.2MB

So let’s try 800%. Here is the file size before cropping.

Here is Gimp’s output cropped to 1920×1080.

Upscayl has a feature called “Double Upscayl” which runs the upscayl twice on the image. Running Upscayl takes longer that Gimp, especially if doing the double upscale. Here are it’s results. So first I left the scale factor at 4x and checked the “Double Upscayl” checkbox. This resulted in a much larger image, since it first did a 4x, and then another 4x. That is a 1600% scale factor. That isn’t what I wanted, but here is that result. First the uncropped image size.

And the cropped image. At this level of zoom you can see that the AI has sharpened the feathers too much.

Ok, now lets run Upscayl at 8x, without the doubling checkbox checked.

Here is the uncropped image size, it matches the 800% Gimp scaled image.

So what do I think? Well it certainly looks promising. Upscayl has different algorithms you can choose. I was using the ultrasharp. I also suspect that the image subject will make a difference. Does the image have blurry areas; how does it do with people; what about text in an image, etc.

Posted on

Faro Simulator as a stand-alone Windows executable

Someone asked me if I could code the faro simulator as a stand-alone windows app, so they could use it without being on-line. So I did 🙂

You can download a zip file containing the file faroSetup.exe at http://robertjwallace.com/faro/faroSetup.zip

Download the zip and unzip the file to get the installer. Run the installer. It should work, but use it at your own risk. It doesn’t access anything on your computer or track anything. Note that when you run the installer you may get a “Windows protected your PC” message. That is just telling you that Microsoft sees me as an unknown publisher. If you get the message, click the more info link and then click the run anyway button.

Posted on

A wall mounted control panel for home assistant

While not the most glamorous, or largest, or most expensive control panel, this meets my needs perfectly. I wanted a small control panel, using Home Assistant, mounted by my front door that I could use for controlling my garage door, to turn off all the lights, etc. I wasn’t interested in spending any money for a tablet to mount or a touch screen. My solution was to re-purpose an old Android phone I had laying around. By having this right by the door I don’t have to pull my phone out of my pocket and start up Home Assistant to do the set of things I want to do leaving or entering the house.

Continue reading “A wall mounted control panel for home assistant”
Posted on

Faro coincidence

Boy the world has changed! When I was young if you wanted to learn magic you had to hang out at a magic store. A real brick and mortar building. To get to the real good stuff you had to make friends with the magician behind the counter and prove your worth. If you were lucky they might show you something more than “Scotch and Soda”. They might recommend a really good hardback book on magic, but those books weren’t cheap. You had to pay your dues, so to speak.

Today you have the Internet. The brick and mortar stores are almost completely gone. I miss them.

Anyway, while I was surfing the web I found this self-working coincidence trick that looked interesting. Now like a lot of these tricks on the Internet, this one had several “kickers” to the coincidence effect. Go ahead and take a moment to look at it. It is at https://youtu.be/9KhQrR5uqN8?si=2rnxeBWteVBaXC7m .

Personally I don’t like most “kickers”. They detract from the first effect. Also I prefer tricks that don’t require a setup. I want to pick up a deck, let the spectator shuffle it and then amaze them with an effect.

So this is my version of the coincidence trick based on the method used in the YouTube video.

Method

Have the spectator shuffle and cut the cards.

Take the deck back and using Marlo’s Faro check, spot the 25th card down from the top. Simple split the deck for a perfect Faro and begin the shuffle to see if you cut it exactly in the middle. Then un-weave the shuffle and spot the 25th card as you place the top half back on top. This is done causally.

Now do two Ireland shuffles (explained in the video). These keep the 25th card in place. As you perform these shuffles patter about coincidences.

Dribble the cards from the right hand to the left and stop someplace after the middle of the deck. Throw the card stopped at to the table.

Fan the deck with the faces towards you and place your card, reversed, into the deck to the right of the 25th card, which you spotted at the beginning. Close up the fan.

I then do some sort of false cut, keeping the deck in order. I use the Jay Ose False Cut because it is simple and very clean looking. There are more complicated cuts such as up-the-ladder, but for this effect at this moment in the effect I don’t want to look like a professional gambler.

Dribble the cards again, having the spectator stop you before you reach the middle of the deck. Drop the top block of cards, from your right hand, to the table and have the spectator take the top card from the left hand block. They look at it and place it on top of the tabled portion. Drop the cards from the left hand onto the cards on the table, “losing” their selection. (At this point the two selected cards are exactly 26 cards apart.)

Have the spectator give the deck a number of single cuts.

Pick up the deck and do a perfect Faro shuffle. It can be either an In or an Out Faro. This shuffle places the two selections next to each other.

There should be a natural break where the face up selection is located. Cut the deck a little bit below this location moving the two selections to near the bottom of the pack.

Deal the cards into two piles, alternating. This brings the two selections to near the top of each half of the deck.

You are now ready to show the coincidence. Take cards off of each half, one at a time until you reach the face up card. Ask the spectator to name their selection, and turn it face up to show the coincidence.

Additional notes

If you can’t do a Faro shuffle you can still to the effect. Take the shuffled deck back from the spectator and turn it face up in your hands. Show the deck is mixed by spreading groups of five cards from the left hand to the right, and dropping the group of cards to the table, in a pile.

Do this four times (dropping 20 cards) and then spread six cards and drop them to the table, remembering the top face up card. Then just continue to spread off groups of cards, dropping them onto the pile.

Pick up the cards and turn them face down. The card you remember is at the 25th position.

Have the two selections made as describe above. Let the spectator give the deck single cuts. Throw in a couple of Charlier false shuffles. Now make two piles of 26 by spreading the cards in groups of two or three and counting them. This is like in the video. Now finish the effect as above.

Posted on

Another Lie Detector

This is another lie detector plot. I have another that I also perform which you can read about at https://robertjwallace.com/lie-detector/. This one is easier.

Start by having a card selected and controlled to the bottom of the deck. The convincing control is a good method which you can learn here: https://www.youtube.com/watch?v=78aZh6eUneI&pp=ygUSY29udmluY2luZyBjb250cm9s

False shuffle and/or cut keeping the card on the bottom.

Set the deck down and ask the spectator to cut around two thirds of the cards to the side, to eliminate them. This cut is key. The bottom part of the deck must contain at least 15 cards, but not more than 23.

Pick up the remaining cards and explain that you are going to use the deck to determine the card by asking the spectator to name their card, but allowing the spectator to either lie and name a different card, or to tell the truth and name their card.

Demonstrate by dealing out the cards, spelling the phrase “your card is the”, dealing one card for each letter face down in a pile. Drop the remaining cards on top of those and ask the spectator what their card was. They can lie or tell the truth. Pick up the pile of cards and spell the name of the card they named, one card per letter.

For example if they say the “eight of clubs” you would deal out “e”, “i”…”s” with one card per letter down to a pile, then drop the remaining cards on top.

Now pick up the pile and deal the cards in the same manner, spelling out the phrase “did you lie or not” (14 letters). Turn over the last card dealt (the “t”), and it will be their card, showing whether or not they lied.

So why does the pile need to be between 15 and n cards? Well this is because of the number of letters in the names of the cards. Cards can be spelled out with from 10 to 15 letters. So to spell out the longest card name without running out of cards you need at least 15 cards. Now after the first deal of “your card is the”, the position of the chosen card is equal to the number of cards in the packet minus the number of letters in “your card is the”. The number of letters is 13. That means the chosen card will be at the 2nd from the top if the packet has 15 cards, for example. For the trick to work, that position must be within the number of letters in the name of the card that the spectator names. This is so that the chosen card is dealt by on the second deal. Since the spectator’s smallest name contains only 10 letters, the chosen card must be within the top ten cards after the first deal. The largest number of cards for that to happen is 23 (23 – 13 = 10).

To visualize this algebraically:

n = number of cards in the packet

c = the location of the chosen card

At the start with c at the bottom of the packet, c = n.

After the first deal of “your card is the”, c = n – 13.

After the second deal, c = n – c + 1. Since, from the last step, c = n – 13, this is the same as c = n – ( n – 13) + 1. This will always equal 14 for n = 15 to 23.

Then after the third deal of “did you lie or not”, c with be the last card dealt.

Posted on

A slightly modified Erdnase Stack

The one aspect of the Erdnase Stack that I have never liked is the undercut and throw during the second part of the stack. Below is the description of the four card stack for a five handed game with that step in bold:

FOR any game in which cards are dealt singly. Three of the desired cards are placed on top, one on bottom. Under-cut about one-third deck, injog top card, run two less than twice number players, out-jog and shuffle off to last card, so that it will be left on top. Under-cut to out-jog, forming break at in-jog, run one less than number players, throw to break, run one, in-jog running one less than twice number players, out-jog and shuffle off. Under-cut to in-jog and throw on top. Under-cut to out-jog, run one less than number players and throw balance on top. This gives the four desired cards to the dealer in four rounds.

Erdnase, Expert at the card table

Normally during an overhand shuffle you don’t simply cut a block off and throw it to the top. Instead, what I do is what was done in the first halve of the stack, namely, undercut to the out-jog and form a break at the in-jog. Then as I start the second part of the shuffle, I let the cards below the in-jog drop onto the top of the deck as I run one less than the number of players. This effectively hides the throw and reduces the right hands shuffling motion.

I think it makes the shuffle more deceptive.

Posted on

Faro stack for 3, 4, 5, 6, 7, or 8 hands, four aces

This was a “problem” I gave myself. Obviously you can stack four aces for a four handed game with two faro shuffles, but what if you want to deal a five handed game. Using my faro shuffle simulator I found a pattern that once the aces are in the correct locations a single faro out shuffle will stack them. (I used my faro shuffle simulator to work out the pattern)

Continue reading “Faro stack for 3, 4, 5, 6, 7, or 8 hands, four aces”
Copyright (c) 2023, Robert J Wallace