12 – RF Laser Machine: Dotting Impossible, So What Now?

0:00well hello again um i’m glad you can join me for the continuing exploration of this

0:06little tangerine tiger last time we reached a sort of a slightly disappointing conclusion that

0:12maybe this machine might not be able to do what i call proper photograph engraving

0:19now i think for the benefit of those people that have not been following my previous session in great detail

0:25i really ought to explain what proper photo engraving is in my terms i have transposed this

0:32picture into a resolution that should suit this machine it’s 254 pixels per inch

0:40which means that every pixel in that picture is point one of a millimeter square let’s just go

0:47and have a look somewhere nice and contrasty and i’ll show you what i mean there we

0:52go look we zoom in on her eye and you can clearly see the square

0:58pixels okay now those are rather strange pixels

1:05because the although this the pixels that you see are square here

1:12there are further pixels on this screen which means i get shades of color within a pixel

1:19so this is not a true 254 pixels per inch image when it’s in

1:26color now we can’t print color on this machine the best that we can do is some sort of brown on organic materials

1:33normally so we’ve got to go through a process of converting this

1:38from color to first of all let’s take the color out of it

1:44and we do that by converting it to grayscale and there we go so we’ve now taken all

1:50the color out but we’ve not changed anything as you can see we’ve still got the same number of pixels there and the pixels are graduated as well but

1:57even so we can’t print various shades of grey on this machine in the mode that we’re currently

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

2:04operating in there is a special operating mode for this machine called grayscale mode where we can

2:12try and emulate each one of these pixels with a different power and we should get onto that later but at

2:18the moment we’re trying to do simple dot for pixel photo engraving so let’s go one stage further now and

2:25let’s convert that image into a bitmap but we’ve already got 254

2:30ppi pixels per inch set for this image okay so now we’re going to do something

2:36called a diffusion dither now there’s all sorts of dither choices trust me diffusion dither is the one you

2:41want and hey we have now converted that image into just one color

2:48black and it looks like two colors on there black and white but hey we’ve got a background color

2:54that we’re going to use that might be wood which is a light brown it could be paper which is white

3:00white represents the background color and black represents the burns that we’re going to

3:06put onto the paper every time we scan across one of these lines the computer will switch on for

3:13black and off for white that was what i was attempting to do in the last session

3:19i’m trying to get the cleanest possible black dot you might say well that’s a

3:25pretty rubbish image it’s only rubbish because we’re looking at it so closely if i start zooming out on that image

3:32again you’ll see that your eye

3:38is gradually and now we’ve got something called

3:43aliasing between the screen pixels and the pixels in my image

3:49let’s just carry on and all of a sudden you’ll see we should get to there we go look we’ve

3:54got to a point where the pixels on the screen and the pixels in the picture match and we’ve got a superb image how did we get to that

4:00lovely grayscale image that you can see that looks exactly the same

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

4:06as the original photograph well the answer is your brain is able to mix the black

4:14and the white that’s on that palette into a gray it’s doing it for you

4:21you cannot see the black dots or the white dots but what you see is an average of look

4:28this density here and this density here is completely different when you can’t see the dots at

4:33this sort of detail your brain is actually creating its own mixing of these black and white ratios

4:42and creating its own little grayscale picture this process is nothing to do with laser machines

4:49this process was developed for the black and white ink printing industry

4:54we’ve hijacked it and are trying to use it for our laser machines but it doesn’t mean to say we can change

5:01the rules basically what we’re trying to do is to produce one dot equals one black pixel

5:09and if we mess around with the ratio of black and white we’re going to mess

5:15around with the quality or the color grading on the picture itself this area here relies on the ratio of black

5:22and white to create the gray that you see in the image [Music]

5:28here we’ve got a single black pixel a joined to another single black

5:34pixel adjoined to another one now if i make that pixel wider than

5:42what’s on the screen i lose the white and if i lose the white i lose the ratio of black to white

5:51so i’m not going to get a true representation of this image if i start making

5:58more black in the picture than white so here we’ve got some of those pixels that we’ve just been talking about

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

6:04and the ratio between the pixels is exactly the same as that on the picture that’s one pixel that says pixel space

6:11pixel space pixel space so we’ve got 0.1 pixels and 0.1 gaps

6:17okay now our i relies on mixing the ratio between the white

6:22and the black of the pixel or the color of the pixel now these pixels will be black right but i’ve left them

6:28white for the purpose of clarity on here when the controller comes along and says okay the machine lace it off later off

6:36later off later off at that point there it will switch the laser on because this is a black pixel and black

6:42means switch the tube on so it will switch the tube on and then it will come to white and

6:48says switch off and switch on switch off

6:54switch on etc now the problem is it is switching the tube on there we’ve

7:00got a beam which is a certain diameter and that beam will produce a line

7:09like that now that’s fine while we’ve got a line but if we’ve got a 0.1 beam and a 0.1

7:18pixel when we switch the beam on the beam will switch on

7:23like that and when we switch off at this end it’ll

7:28switch off like this so we have not got a pixel’s

7:35worth of burn we’ve actually got two pixels worth of burn because we’ve got half a pixel hanging

7:41out the end each end and so this half a pixel here and this half a pixel here and this half

7:48a pixel here and this half a pixel here look they will join up like this and

7:53we’ve got virtually no white between our pixels the laser tube is doing exactly what

7:59it’s been told to do switch on and switch off and because it is so good

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

8:04and rapid at switching on and off this is an insoluble problem it would appear unless we could

8:11find a way of doing that so if we could switch the beam on for a very very short period of time

8:17say for instance one of those we should get slightly elongated the relationship between the pixel

8:24and the gap is going to be maintained so we attempted that last time there

8:31wasn’t sufficient power in that very short duration which was about 20 20 microseconds

8:36to produce a good enough single dot so in frustration i said right

8:43forget the rules let’s just go and do some blind guessing we had to go

8:49doing a picture with various parameters so that i could see what the general

8:55pattern of change of density of the picture was with various factors all of these were

9:03five kilohertz and they were 50 percent 95 percent

9:08and 70 percent when we were only on for 50 of the time we got this sort of picture

9:16when we run for 95 say 100 we’re into this sort of pattern here

9:23where we’ve got the power on for the whole length of the pixel then we got this image 70 well it’s a little lighter

9:31okay so then we changed the frequency let’s compare that one look 95 percent

9:37there at 95 percent here but here it was 5 kilohertz

9:43and here we’ve got 10 kilohertz i’m slightly puzzled about why there’s a difference between these two this looks very muddy

9:49this one looks less muddy you might think that these are quite good pictures some of these but in fact in reality they’re

9:54they’re not very good quality at all now in trying to unravel what’s going on between these two

10:00pictures the only way we can find that out is to look at how the power is being

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

10:05commanded to hit the paper so here’s the signal that’s driving our five kilohertz picture

10:13that’s not quite what i was expecting to see all high all high five percent low

10:20five percent low all hot so we’ve not got uniform power okay it’s only five

10:26percent loss now it is a problem for me because what i’m expecting to see

10:31is every time we start a pulse we reset the time base and we get exactly the same picture well

10:38that theory has been shot out of the water now hasn’t it look because we haven’t got that what it looks as though we’ve got

10:44is a time base running in the background and we are switching in and out of that

10:50time base based on when we require our power so if i make

10:56the time base match the frequency of the pixels then i will get uniform pattern but in

11:03this instance i’ve chosen a bit of an odd number 5005

11:08and it’s caused a bit of a problem as you can see now i suspect that if i chose a nice round number like

11:15maybe 10 000 and let’s just see what the difference between that 5000 signal is

11:21and the 10 000 signal leave everything the same and we’ll change this signal up to 10

11:27okay compare that 5000 signal with the 10 000 signal

11:32it’s showing the same sort of characteristics isn’t it look they’re not in the same place every time we’re not resetting the time base for every

11:39pulse we’re running fast enough to make sure that we always get one piece of missing data

11:46for every pulse regardless of where it is so we’ve got uniform power per pixel

11:54at 10k the difference in the quality of the picture now there’s the difference between the

11:5910 and the 20 kilohertz this one might be slightly darker just

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

12:05might be the only difference i can see is the clarity in the eye there’s slightly better

12:12definition in the detail of that eye than there is in that one let’s go and have a look and see what

12:18a 20 kilohertz signal looks like we’ve hit on a number which creates a uniform look at the time

12:25base that’s running in the background let’s have a look see what’s happening with our fifty percent picture

12:34i think we’re gonna have to forget this first one because it does look as though we’ve got uniform patterns on all the other pixels

12:43but that pattern is not what i expected i’ve got 30 watts 30 watts and 30 watts

12:50and i’ve got 0 watts and 0 watts hang on that’s not 50 average

12:57that’s 60 high and 40 low we’re not working with averages here but

13:02if we want to look at it in terms of average power per pixel we’ve got 18 watts not 15 watts

13:09per pixel now my brain is really beginning to hurt i need a little cup of coffee sorry i need a big cup of coffee and

13:16some time to have a quick think about this i’ve spent some time in a nice dark room

13:23quiet big bar of chocolate two cups of coffee and

13:31i feel like a new man i don’t look like one but i feel like one and i think we might have a little bit of a plan

13:37now we’ve concluded that this time base is operating in the background and all we’re doing is we’re switching in

13:42and out of whatever that time base is now we’ve already concluded that our dot hanging over the edge of

13:49the pixel at each end is going to block the white space in between the pixels and that’s

13:54going to give us the wrong black white mixing ratio when we look at it so it’s always going to create a

13:59darker picture than it should be we can’t do anything to stop half a pixel hanging

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

14:04outside the start point of where the pulse starts but if the pulse

14:11finishes early for example it finishes 50 percent along

14:16the pixel then that means the remaining 50 percent

14:22will be there and so we shall only lose half as much of the white that i’m sure

14:29will give a better quality picture the question is how do we make sure that we stop the pulse

14:37halfway across the pixel this time base operating in the background is the thing that’s going to control

14:44everything okay we’re going to just switch in and out of that time base

14:49now if that is the way that it works and this will prove it beyond any doubt

14:55we should be able to exploit that fact so here we’ve got the situation where we’ve got

15:0150 50 50 high 50 low but the pulse rate is such that we’ve

15:08got three high and two low what we’ve got to do we’ve got to fiddle

15:13with this frequency now so that we get something that looks

15:20like this where that there represents 0.1

15:27of a millimeter for whatever speed we’re running

15:32so this frequency is speed related that’s why we need our

15:37calculation let’s see if we can hunt for that magic number a positive pulse distance of 0.05

15:44and a pitch of 0.1 so i think we’re stuck with having 50 percent

15:50i’m going to push this up now to a thousand millimeters a second and see what that

15:56does wow how about that magic number just appeared look 0.05 which means we get

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

16:04one pulse high one pole slow and point one which means we get

16:11a pixel a gap a pixel a gap fifty percent fifty percent a thousand

16:19one thousand ten kilohertz

16:25uh we have got exactly what i expected we’ve got one third high

16:31and two thirds low so what we’ve got we’ve got half a pixel high

16:36half a pixel low one complete pixel low and then half a pixel high

16:43so this is working exactly as i’m expecting let me go back and explain what i’m

16:50trying to do now so we’ve been able to achieve now half a dot outside

16:56the pixel a scan line of half a pixel now so that now when we do the next pixel

17:02like this we’ve got a gap of half a pixel between

17:08the pixels now that’s not still the correct ratio but it should give us a much better

17:14quality picture than we’re seeing here before i go and do that look what i’ve got here

17:23i’ve got a piece of glass which has got a really nice hard crisp surface

17:28we’re actually going to try those settings on a piece of glass and see what they come out like under the microscope because

17:34this should give us a fairly clear impression of whether or not we’ve got the right

17:40sort of pulses i think we’ve got a result this time give or take a little bit

17:46we’ve got a 0.1 dot and a 0.1 gap

17:52not perfect and you can clearly see i think the start of each pulse here

17:59with a with a splash the other thing that’s interesting as well we’ve got perforated continuous lines as

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

18:04well we’re more interested in these single dots although perforated lines up here

18:10might possibly give us white where we don’t want white so i think the only way to really test

18:15this is to go back to my picture and use these settings and see what

18:20happens i don’t think there’s any doubt that this is a substantially better picture

18:26than this one it’s less muddy look at the clarity of

18:31the eyes this piece of hair here is less fuzzy you can see all the hairs up here you

18:38can’t see them here so that’s a definite step forward

18:43we can’t get any more color into that picture unless we slow it down there’s nothing

18:50else that we can do i’m pretty confident the other reason why that picture is very light

18:56is because they’re going to be short sausages now but they’re still dashes but those sausages will not be 0.1

19:03wide because we’re running so fast the actual lines will be probably something like about

19:080.07 or 0.06 as we found out when we measured the lines on the test

19:14previously so we’ll go back and we just measure our lines on here and see what size the dots

19:19are because the other reason it could be light is because what we’ve got we’ve got sausages

19:28that are like that with 0.1 pitch in other words we’ve got white between the sausages

19:35so that’s yet another reason why we’ve got a light picture well i think the reason why it’s

19:41so like that picture when we look at this now is the variation in the burn

19:46look we’ve got little teeny weeny spots right in the center of some of these and yet look these are continuous lines that

19:53should be brown and they’re very light straw so we’ve not got uniform burning we can

20:00overcome that problem quite easily by slowing it down but hang about slowing

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

20:06it down changes the ratios so let’s have a quick look

20:11to see what we can do so if we slow it down to say 500 the pitch is wrong

20:18can i put that up to 20 000 and correct it you’ve gone the wrong way yeah i thought so so i’ve got come back

20:24to 5000 to get the same result but darker okay now i can’t use 5000

20:30so i’ve got to use probably something like about thousand and five five hundred speed fifty

20:37fifty five point zero zero five as close to five thousand as

20:43i’d probably dare get okay now here’s a little tip whenever i run a program and i press start now as soon as the

20:50head starts to move i press pause and i wait until the program is downloaded then i

20:58allow the program to proceed i have found on occasions that the machine can’t walk and talk at

21:04the same time it needs to receive its program before it carries on sometimes it will lose its

21:10zero and drift all over the place

21:25[Music] well i think the point is well proven

21:30here look five kilohertz five kilohertz

21:38800 millimeters a second 500 millimeters a second so this should be substantially

21:43darker than that one but then again this one is only 50 percent

21:49and this one’s 95 but look at the difference in the quality of the picture

21:57i mean the quickest and easiest way to just look at the eyes the detail in the eyes

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

22:02the subtleness of this shadow down here it’s not existing the face is blotchy this is it’s not

22:09over burnt whereas i’d say that this to be honest is probably if i give this a bit of a rub here

22:19look slightly over burnt which means the pixels are overlapping

22:26just here not too bad but definitely down here in this dark area so we’ve still got a bit too much power

22:33there so that gives us the possibility of maybe going up to maybe 700 millimeters a second lighten it a

22:40little bit we’ve got no other way of playing with this picture now except speed as we said right from the

22:46beginning there are very serious limitations on what we can do there are very complicated combinations

22:53of things that affect the quality of this picture the most important one is the one that nobody

23:00ever talks about which is this pixel overlap while i was in my dark room consuming

23:06cups of coffee and enjoying bars of chocolate or a bar of chocolate sorry i had another thought which i’ve got to

23:13experiment with that involves messing around with the picture itself so here we are in photoshop with the

23:19original image which is 300 pixels per inch 52 centimeters half a meter wide

23:27and 78 tall so we’re going to take this picture and we’re going to adapt it now at the

23:35moment if i was to zoom in on this picture you can hardly see the pixels here

23:44300 pixels per inch but you’ll notice that the pixels are square which is what we’ve always

23:51said they should be because we’re trying to copy square pixels with round dots we’ve got to go through all

23:58sorts of acrobatics to try and get even close to that square and we shall never be able to do

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

24:04it with this machine so we’ve got to admit failure from that point of view this machine is not capable of producing dots

24:11it might be able to get close to dots but the closer it gets to dots the weaker the image becomes the less

24:18power you can put into a dot we’re going to change our strategy now

24:23we’re going to create a new picture file new and this time we’re going to set the width to millimeters 100 millimeters 130

24:32tall but this time we’re going to have 254 pixels which is the 0.1

24:37pitch between the lines and across so they’re square pixels but now

24:44look here it says it pixel expert ratio square now this is something i’ve never played with before because i’ve never needed to

24:51but i did remember that there was something here which might possibly help me with my

24:58problem we’ve got all sorts of options here we can play with look from one all the way up to two so let’s try this

25:05one here at the extreme to start with two to one okay

25:11so now i’m going to pick up my image from here and i’m going to drag it

25:16across into here and there’s my image hold down the shift key and shrink it to the

25:22approximate size that i want okay so i can’t because the ratio is

25:28wrong um maybe i ought to start off with a different ratio then i don’t know

25:34so we start again and we’ll try and correct for it from the start so we go new and this time we’ll instead of having

25:40100 we’ll put it in at 50. so that looks to be about the right aspect ratio now

25:46so if we drag this picture across into here control minus

25:52and drag a corner

26:00we should be able to make the image roughly the right sort of size and then we’ll drag it onto the window fit on screen and we can see what’s

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

26:06going on so we put that to the top of the screen now we should be able to do view fit on

26:11screen and it’ll come a little bit bigger there we go so look there’s no difference between the quality of those two pictures

26:17so it would appear what have we done to this picture now so let’s zoom in to

26:25her eye again and what do you notice now about the pixels

26:32they’re not square they’re rectangles so how does that help us are you ahead

26:38of me without doing anything to the image at all we’ll just change this to grayscale which we have to do first and

26:44then we’ll go image mode bitmap the bitmap should already be 254

26:50which it is 0.1 pixels per inch now look what we’ve done we’ve no longer

26:55got square pixels now how does that help me well what it

27:00does it increases the amount of white between the pixels in the horizontal

27:07direction if we’ve got half a dot hanging over the end of each one of these black pieces now we’ve got a whole pixel in between

27:14which is white now has that changed the relationship well it must have done because we’re no

27:19longer blocking out all of the white with our pixel overlap and we should be able to get even better

27:26than that if we can use the same trick and technique that we’ve just demonstrated of

27:32setting it up in such a way that we only use half a pixel so we should be able to

27:39improve the contrast ratio or the mixing black to white ratio even better

27:45okay let’s see how that comes into rd works [Laughter]

27:51well it’s come in at pixel ratio one to one so we should be able to change this ratio here

27:59from 100 to 200 percent in x

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

28:06and there she is again now if we zoom in and look at it this time yeah look we’ve still got our

28:11rectangular pixels there might be slightly less pixels in there overall so we might lose

28:17just a little bit of detail but hey if we get a nice crisp clear picture i recommend bar of chocolate for

28:24thinking it helps a great deal got nothing to do with the pretty face it’s all to do with the size of the

28:29pixels well the ideal speed that we wanted was a thousand let’s go back to that to start with and

28:3750 50 advanced 10 kilohertz not my dream but getting

28:43close a thousand millimeters a second 254 pixels per inch okay

28:53it’s definitely not as good it’s lots of detail probably because of the pixels or is it

29:00because of the settings that i’ve got see i’ve changed the pixel ratio now perhaps we want to go back and look at

29:06the calculation i think is the correct term it’s nothing near as good as that one

29:12now i suspect it’s probably because i was a bit too anxious and quick and i didn’t really come back here and

29:19look at what we should have been using we just compared previous settings now the previous settings worked well

29:26but now look we’ve got point one pitch whereas in fact the

29:31pitch is now 0.2 because that’s the length of the pixel

29:370.2 so here’s the situation we now find itself in we’ve got rectangular pixels with rectangular

29:44spaces between them we’ve got a dot on this end of the pixel and as we travel along half a pixel

29:51and finish we should produce another half a dot okay so now we’ve got a big black space

29:59here if i was to do that i’ve almost created

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

30:05a complete pixel by doing this as you can see i’ve lost a percentage of my black but i

30:11haven’t lost any percentage of white because effectively every one of these will shift along and they will all be

30:18the same pitch i have a 0.2 pixel and a 0.2 gap a 0.2 pixel and a 0.2 gap

30:24so this has effectively fixed our problem we haven’t got any more hanging out

30:30dots the dots are now sitting inside 2 pixels but we’ve got two pixels spaced

30:37between each dot so we’re back to square one except that we are have got a different

30:44resolution in x at probably about point two as opposed to y where

30:50it’s still 0.1 should give us good results so what are we looking for well we’re

30:55looking for a pitch of 0.2 and a pulse of 0.1

31:01what sort of magic numbers give us that result can i go back to a thousand delete and say enter and then

31:09this is still going to be 50 and let’s make that

31:1410 000 10k does that help me um

31:22no what happens if i take that up to 20k that looks possible no going the wrong

31:28way that means i’ve got to come down to 5000. now at 5000 i’ve got my 0.1

31:34pitch and my 0.2 distance between the pixels or pixel

31:40length so that’s pixel length and that’s half a pixel length because we’re on 50 percent

31:46thousand millimeters a second and five thousand and five

31:54now this approach is not the same as that adopted by the

32:01so-called big boys their software works with a different mechanism they’re

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

32:07effectively drawing different length lines but they’re not drawing dots they create let’s call them blobs

32:14from lots and lots of dots so they don’t attempt to print dots what they attempt to print

32:20is lines by making a fairly coarse picture of blobs with like a newsprint

32:27style image now the result is always a bit cartoon

32:32looking rather than photo looking and now i can see why we’re having to go

32:38through all these mathematical gymnastics to find a certain combination that suits what this machine is capable

32:46of doing so in a strange way it’s madness this machine is just not really suited to proper

32:54photo engraving sad as it seems we might be able to get some half decent results out of it but

33:00they’re compromised results they’re not 100 dot for pixels

33:14right well with this picture we’re going out of the so-called comfort

33:20zone for the machine some people say i can work right down to zero

33:26for frequency and other people say well you shouldn’t go into the pre-ionization

33:32zone which is sub 5000

33:38or 5 kilohertz well i’m not going to apologize because one of my rules is to break the rules

33:45and here i am working at two and a half

33:50kilohertz to try and get some color into the picture because i’ve reduced the speed to 500

33:58now in reducing the speed to 500 and still satisfying my equation of half a pixel

34:08i’ve had to go down to two and a half thousand kilohertz

34:19seems as i’m going to get rubbish doesn’t it i should have listened to all those

34:32people

34:46i have to say that of all the images that i’ve done that one takes the prize that i can

34:51compare it with i mean there’s nothing on my original sheet that comes anywhere near

34:59some of my earlier attempts today

35:04no no that’s the closest one

35:10this one has got a degree of crispness to it but it’s over burnt and when we try and correct the over

35:16burning we get nowhere this one has got a lovely soft

35:22almost a matte photograph look to it if you didn’t know it you’d

35:28think this was part of the studio out of focus because just the face is

35:33the focal plane where we’ve got those lovely crisp eyes the teeth and the lips and the hair and everything else is

35:40dropping out of focus i think we’ve got to be moderately impressed with that

35:54i would say that comes 90 or more as good as i could do

36:01on the glass tube laser so I’m pretty pleased but having said that

Transcript for RF Laser Machine: Dotting Impossible, So What Now? (Cont…)

36:10i had to run at 500 millimeters a second which is nowhere near my target so i think we’ve

36:15got a little bit more work to do but it does show potential promise we found some wiggle room

36:21and some techniques that we could possibly use to get some half decent photos off of

36:27this machine now whether we can do it for other materials we might have to investigate that next time and see what other

36:34tricks we can pull and hope that this has been an interesting

36:40session for you it’s just another stepping stone to understanding what this machine can do so until the next session

36:47thanks for your time and I’ll catch up with you then