Forum › Posts by MrEngenious
This isn't the place for personal quabbles with each other's gaming habits nor is this image thread the place for Genshin Impact. I will say that I recommend not equating someone's dislike of a game to an attack on your person. If you enjoy a game then enjoy the game; you don't need other people's input to continue enjoying it. Gacha games will get tons of dislikes fairly and unfairly that is just what it means to be a major gacha game.
@thechampionmike95 The argument was inherently about bisexuality, without anyone using the term, even if the image wasn't tagged as such. I've added the tag now as I had no idea about the context of the series to confidently add or remove tags prior as tagging is handled by only so many.
last edited at Nov 30, 2024 10:19PM
Pruned some comments and handed out some temporary removals. This yuri purity shit because bisexuals exist is not welcome.
last edited at Nov 30, 2024 9:34PM
This argument has gone on long enough. People are allowed to enjoy "crumbs", whether old or young. For all this talk about not being anti-bi, y'all sure are keen on making sure that it can't be classified as real yuri in any context because they may also like a dude sometimes while liking each other too. I'm sure it's unsatisfying for y'all that it's not just women in the series but this purist shit is incredibly annoying.
There will be no more talk in that regards. Shit on it being from a harem series all you want, there will be no bi-erasure here.
last edited at Nov 30, 2024 3:29PM
You are all free to have your opinions, however being anti-bi is not going to be one of them, okay?
@ShortHairedCrow:
Your post was fine for the first half, then you kinda went off the rocker with the second half. There was nothing Drago said to insinuate they were part of this nefarious group that wanted to a reboot of the series for the wrong reasons. Focus on the argument you wish to make without using logical fallacies to attack the person. I have cut down the offending half of your post.
@Drago928:
I have removed your edit.
People are free to not like certain ships. As a general guideline if you're going to like/dislike a ship and make public posts about it, add to the conversation with reasoning from the source material rather than attacking the posters sailing the ship or sinking it.
thechampionmike95 has not posted that "toxic shipping should be outlawed" but rather that the background of these characters make it problematic for them to enjoy.
Ships should be fun and crack ships spicy fun. If you're going to get angry at a poster for not liking a ship you like (because of canon violence/trauma, no less) and attack them directly then I would suggest you re-evaluate your priorities.
I do suggest everyone be less aggressive in general here.
While no one has broken any rules explicitly, this conversation has gotten a bit deeper than the image topic section was accommodating for.
I'm going to have to request that this specific line of argumentation conclude here and either move elsewhere, like a dedicated post on the forums or just perhaps not trying to bring twitter, formerly known as a useful website, to the forums any more than it has.
And while the background information is relevant to the image in question, it has clearly overtaken the focus here and moved into political considerations of all people involved.
Posters have already made clear why such art by those mentioned and unmentioned artists are harmful rather than something one can completely ignore as "artistic license" when it's being actively shared with malicious purpose. While we are not outright banning opposing viewpoints on this matter, we are going to clamp down on it when to goes from curiosity/questioning to staunch defense. If people want to live and let live, then let the gays/trans/enby live too.
Comments for this piece cleansed. Didn't want to delete a few posts, but they wouldn't make sense without the posts I wanted to remove. Please do not bring imaginary drama from a dying platform over into here. If there's something you wish to critique, then keep it thoughtful and constructive.
I found a pile by accident on my drive. Doesn't feel like some of these were ever "official" so some of these may be contestant submissions.
last edited at Mar 16, 2022 1:40AM
There's several instances here that require redrawing on this page that 03_177 did not test for at all. I would absolutely recommend whiting out all of the text after you've performed all the standard page-wide adjustments. Removing the text will allow you a bit of a clearer view of what needs to be done to fill in the newfound gaps. You'll run into some difficulty with the very last bubble so I'd leave that alone if you're following along and check out how I chose to handle it.
One tip I can give is when you're whiting out any text that is touching the bubble/panel borders is to carefully use a smaller brush to clean away the text while leaving the contours of the bubble/panel intact.
It might take a few tries, or even some repeated brushing, to get the "right" movement of the brush. Just have your CTRL + Z/CTRLZ + ALT + Z ready to try over and over as needed.
This is how I made my look. If you have trouble making the border lines look smooth make use of the blur tool at 5% or less strength. Using the brush tool with low or 0% hardness could also help.
The last bubble is a little trickier to gauge what is a proper border to leave it with. You can either take your time with it or remove the bubble bits altogether and redraw the bubble lines instead.
Let's begin redrawing the easier stuff. Starting with the second panel we have a bubble and a bit of clothing that needs to be redrawn. This will make a simple case for the pen tool and the clone stamp tool, although if you have actual drawing skills/tablet then you may be able to just draw it in manually or using your mouse if you feel confident enough.
The way the pen tool works is that you create anchor points for the path to swerve towards (incredibly sharp corners may not work well). The path's curves can be adjusted by using the tool to grab ahold one of the arms associated with the anchor point and moving it around. Click on a different anchor point to change the angles associated with that point. Most of the time my redraws focus on two points, although there have been a few rare occasions that using more than two worked out for the redraw.
The path itself doesn't do anything since it's only meant to be a guide for what you intend to do with it next. What I do all the time is perform a stroke using my currently set brush settings by right-clicking, choosing "Stroke path...", and then selecting the brush tool. I set my brush to a width slightly larger than the line I'm trying to imitate and do 1-2 strokes using the pen tool. This will create a wider, but (hopefully) properly curved line that I will then use my brush to trim down to the appropriate size and look. I do it this way because the lines from the scanned image do not look perfect like the brush tool from photoshop so I have to purposefully cut down my redrawn line to look like it fits in. This includes some use of the burn/dodge/blur tools to soften the redrawn line I made.
You may have noticed that there are colors at the bottom of the right-click menu. They're there purely for visual organization on top of the use of folders. So for instance you could set up a personal system where sfx text are red labeled and dialogue text is blue labeled. I never used it but I can see its usefulness for quickly finding things.
If you'd rather try to brush properly or do some other ways of drawing a simple curved line then by all means go for it! I don’t have a drawing tablet so I’ve found that the pen tool is quite versatile for helping me do these redraws.
Also be sure to draw lines on a layer above the leveling/curves adjustment layer. Unless you want to adjust your choice of gray to fit in with the adjustment from leveling, it'll be easier drawing with black above the leveling layer and avoid it looking burnt out.
That leaves the clone stamp tool to be explained as there's certainly going to be a lot of its use here. I always zoom in a lot so that the individual pixels are quite large on the screen.
Before I explain this panel, I'll explain how the clone stamp works. By holding ALT and clicking on a point in the image, you set a source or sampling point where when you start brushing with the clone stamp tool it will copy from the sampling point you set. It's copy and paste but in very tweakable brush form.
Unless you plan on doing something a bit more fanciful, I would leave the mode to Normal, Opacity to 100%, Flow to 100%, and Sample either Current & Below or All layers depending on what needs to be copied. Aligned should be ticked on or off as you need it. What aligned does is when you clone stamp, having it ticked on will keep going relative from the sampling point you left the brush off at rather than going back to the sampling point when you brush a new location. I can't quite figure out good wording for this so just play with the tool with it on and off to get a clearer idea. I tick it on and off frequently depending on my needs; the usual mode I start with though is off. And last but not least, turn OFF the icon to the right of Sampling which will allow your clone stamp to accept the adjustment layer into your cloning attempts. There are a few reasons why you would have it turned on (in fact in this image I do have it on) but for this stuff it'll be easier doing these clones above the adjustment and line redrawing layers.
Clone stamp is a vital part of the redrawing process. If you have a more recent version of Photoshop the Content Aware capability is actually okay, although I still won't advocate for its usage too much. It can be handy for getting things "close enough" but without knowing how to do it manually, I'm of the belief that you won't know how to correct or recognize its shortcomings.
You don’t necessarily need to use the Clone Source window, but it can be useful if you want to turn off or adjust the preview it will show on the brush. There’re also more niche things that you can tweak with it that I never truly found a use for but it could work for certain annoying redraws.
Let's start with Sumika's hair in panel 3 after we redraw the hair border:
You may have noticed that her hair has a slight gradient that goes darker as you go down. That means we're going to want to copy roughly on the same y-axis location as the areas we want to fill in. I've already shown where I would personally set my sample point to start with and then brush in an area directly to the right of it where that sampling area would fit. There are patterns within these pixels and for these patterns specifically I usually like to set my sample point on the white pixel in the center of the plus pattern. Set it wherever you can recognize the pattern and continue it if this doesn't work for your own recognition preferences. I’ve drawn some very crude lines signifying the y-levels you should be cloning to avoid off-gradient copying. It’s much easier for people to notice broken patterns than it is to recognize a slight error.
It will certainly take you some time to get used to the idea of this and you will most certainly copy over details you didn't mean to copy. Just keep practicing, shifting sampling points, doing big brush movements or small ones, and it'll become sort of second nature eventually. If you clone too far and have to undo but the spot you started with was good, then tick on "Aligned" and then you can clone it again more easily (and more carefully this time!) since the spot was saved.
There's not a ton of area to work with here so I did have to copy over the same spots more than once. If it's too obvious then you can use the standard assortment of tools to make it more "unique" or take a few bits and pieces from the right-hand side that's harder to work with.
You'll likely find out that the tones are not actually perfectly situated or even (either due to the straightening or the scanning or just how it ended up) and so you'll have to try your best to bullshit it to not look fuzzy. That is legitimate advice when it comes to redrawing as it isn't about making it look perfect it's about making it as unnoticeable as possible. Use CTRL + ALT + 0 to zoom back out to 100% of the page size to see how noticeable it may look and adjust as needed.
I started with Sumika's hair because the gradient shift was more straightforward than the background stuff above her head, but now we have to fill that in. This is a radial gradient so the black is strongest from the corner of the panel and gets weaker as you get away from it. Pick your sampling sources appropriately to match the gradient shift as best you can. If there's not enough area then bullshit it with burn or dodge tools. I didn't need to do much bullshitting as there was a lot of area to work with to get most of it. A little bit of burn/dodge was deployed to make the copy and paste less noticeable.
For panel 5 the pattern is easier. Arguably could've started with this one for introducing the clone stamp tool, but I'm of the type to prefer being thrown into the fire to learn from all the mistakes that can be made. This is straightforward compared to panel 3; pick a starting pixel for replicating the pattern and then go nuts. As there's no gradient on these you can optionally sample from various sources to maintain a more unique set. Just be aware of the times when the tones don't line up 100% due to the scan or rotation.
Panel 6 looks annoying from a distance and is, in fact, annoying. While there are plenty of lines to choose from to clone over they all have their own semi-unique pattern. Don't be afraid to overwrite any part of the original line because remember it's not about perfectly replacing it, it's about bullshitting your way through. If you do a wholesale copy and paste it'll be kind of obvious so I continue using clone stamp and changing sampling areas every so often. You may see that the artist actually has these in a pattern of patterns except that every line is slightly different. If we manage to align with the pattern of patterns properly then we can clone multiple lines at a time rather than one at a time. If you sample from the line itself you find out that you quickly have to copy the same part multiple times and also that the line changes as you go from left to right. Cloning vertically seems the better option.
For panel 7's bubble I would wait until typesetting is finished for the bubble and then fill in the gaps that remain the best I can.
For typesetting, the only new thing is dealing with the sound effects within the bubbles and using new fonts for the non-bubble text. I also said I'll be using new dialogue fonts for every page as well so there will be that too.
For p2f1 I would use a font that is a bit thinner than the dialogue font and maybe a little curvier. I chose KG Love Somebody for the floating text on the first bubble, Nothing You Could Say for the dialogue, and Chinacat for the in-bubble sound effects.
A stroke needs to be added to the text of p2f1 to allow it to mimic the white outline that was present in the original and also to improve legibility. Right-click on the text layer and click on "Blending Options..." or double-click on the text layer itself. The window that pops up shows a bunch of options that can be applied to the layer. I implore you to experiment with all the various options here as I will not be going over a lot of them despite their usefulness.
Tick the stroke checkbox to activate it and then go inside of its options and adjust it to size 3 px, outside position, blend mode normal at 100% opacity, fill type as color, and set the color to white. You’ll see the changes occur in real-time so if it doesn’t look right then by all means adjust it further.
While I don’t do it myself, I would recommend that you use the point text input for sound effects or other text that isn’t standardized since it’s much more flexible to warp and modify compared to the text box. For p2b3 I rotated, warped, and applied the Rise style to the text with a decent amount of negative bend, a decent amount of horizontal distortion, and a minor amount of vertical distortion. This is entirely subjective and I’m not entirely satisfied with the way it looks, but the bubble certainly isn’t interesting enough for me to consider spending more time on it than I already have. It’s better than just rotating the text and calling it a day at least.
In p2b5 I do the same thing except I tweaked the settings and rotation a bit to make it not look like a carbon copy of p2b3. Keep in mind that the boundaries will remain the same if you duplicate the warped text layer so it won’t be viable to add more lines if you didn’t keep space available if you use the box method for inputting text. If you want to just duplicate similar sound effects and save a few clicks from re-doing the settings then the point input will also give you a bit more flexibility if you do need to change it a bit.
For p6b as mentioned I recommended putting the text as desired in first and seeing what needs to be done afterwards. I chose to switch up the font this time to add more of an emphasis to the scream that the previous font couldn’t portray without heavy modification. I used CCSmash this time since it was bolder but was still pretty legible. I took the extra effort for this bubble for typesetting (separate letters scattered around) to save the stress from trying to bullshit the cloning by covering it up. I split up the text into separate layers which introduces a different problem when I apply a 6 px stroke to each one in that the stroke then overlaps on top of each other. This isn’t the only way to resolve it, but the simplest way I handled it was I brushed black on a new layer above all of the text layers to remove the stroke over the letters and connect them.
After you place the text with stroke and any settings you desire, hide the text (or have the text layer set to <50% opacity) to show the bubble itself again. This is another time for bullshitting since there’s not much pattern to clone over and there’s a gradient applied to it. I ended up cloning over the same spots repeatedly to fill space and then adjusted the results to seem less… bad. With the text layer on partial opacity, you can see areas where you don’t have to fill in, although filling them in would allow for somewhat smoother cloning by giving yourself more area to work with.
The key to any redraw isn’t skill, although there’s some of it you will acquire over time, it’s time. There is typically no quick and simple solution (sometimes even with a well-placed content aware) to fill in large gaps of missing area. It took a bunch of cloning, dodging midtones, re-cloning, re-dodging, burning, etc. With a ton of those adjustments, I was able to get something like the final page shown below. I’m not entirely satisfied with the end result since I would probably spend a bit more time had it been for an actual release, but I think it passes a quick read by the average reader. I wasn’t a fan of this bubble before and I certainly am still not a fan now either.
It can be argued that the patterns may be easier to work with before doing the page rotation. I haven’t really bothered checking myself so there’s the potential that cloning patterns may be easier if done before a rotation to straighten.
An alternative is to wholesale replace the entire bubble’s screentone with one you define. The key in defining a pattern is to make it repeat flawlessly so I chose a small segment from panel 3 to copy over.
I hope it’s clear why this is an acceptable pattern to set due to its ability to be copied above, below, and to the sides without causing issue. Go to Edit -> Define Pattern… to create your own custom pattern from this. Then use the marquee/lasso/magic wand/quick selection tools to create selections of the bubble to replace with your pattern. When you’ve selected the area you want to fill, right-click on it and select Fill… and set it to your pattern and see how it looks. You’ll probably have to manually make it look more interesting or at least somewhat more like the original. I wouldn’t quite suggest this option since if you just leave it with the filled in pattern then you’ve essentially made the bubble more boring compared to the original intent.
Now with the final bubble filled in at last I finish the page off by typesetting the floating thought text in panels 3-5. I usually choose a new font to portray thoughts so in this case I went with Shadows into Light. For panel 3 I don’t have the text hover over the head because I feel confident enough that it won’t be noticed that easily. If you’re not confident then you can hover the text over it although you’ll likely cover up more art than was covered up before with the way the horizontal text + stroke will sit on her head and the text will be off center. The translation for panel 3 is one short sentence so I centered it to the panel by widening the text box borders to reach the panel border while setting the text to center align. If there was more room in the panel, I would have let the entire sentence be on one line. It might have fit if I shrunk the text or maybe used a different font.
I center the text for all three panels and have it progressively lower to follow a more “reading” orientation. It also helps to cover up any weird mess that may have arisen when redrawing panel 5.
.
.
.
If you’ve been following this long to get to here, then perhaps you’ve begun to identify potential problem areas upon opening and looking at the raw image for the first time. This will be important if you aim to be a Quality Checker now or later on as you get more familiar with the process.
So for this page I see that the page is pre-leveled, has a decent amount of jpg artifacts from being saved at a lower quality jpg, and panels 1 and 2 have moderately annoying sfx to redraw but can be mostly mitigated. Also, specifically for y’all, there are two new elements to typesetting that are introduced here: heart symbol inline with a sentence and use of bold emphasis in dialogue.
Going through the routine again, you may wonder whether there’s a faster way to do your standard layer setups without having to click a bunch of buttons and re-enter the labels over and over. Open up the Actions window by going to Windows -> Actions (shortcut ALT + F9) and here you can record actions that can help you cut down on some of the repetitive setups, for exporting, and whatever else you may find useful to repeat.
By default all you’ll have is the “Default Actions” portion. Just like with the layers window I decided to make a new group, called it Scanlation, and then created a new action within it called “Default Setup”. When you finish creating the new action Photoshop immediately begins recording your clicks so if you’re not ready then click on the stop button (the recording button will be lit red if it’s actively recording now). Starting with just having the background layer, everything that is described in that action is what I did. Actions labeled “Set current layer” means that I did something like rename the layer. You can see that it even tracks whether I changed my current selected layer, which I took advantage of, to make it finish the action by selecting my cleaning layer.
I chose not to record making the leveling adjustment layer so that I can add it in manually because its requirements change between every page here anyway, which may not be the case if the scans you have are reliably consistent. I made a few mistakes during the recording which needed to be removed so don’t be afraid to stop recording and then delete the erroneous action, although you can also just wait until you record everything you want to do and then clean up the steps afterwards. With these steps saved you can always play the whole action or single out one or several steps to enact as you feel like it by clicking or highlighting and entering.
With the understanding that this page is naturally littered with jpg artifacts, I recommend a few additional steps before leveling. As usual I excessively clean out the bubbles, the empty space inside of the panels, and the white space outside the panels. The rectangular marquee selection tool is handy for quickly whiting out the areas outside the panel borders.
The prevalence of the jpg artifacts means that it’s going to be quite frustrating to manually get rid of it all. The goal of this image for testing was to see whether laziness won out in the end with an overlevel or just completely ignoring the jpg artifacts and thus leaving the entire page dusty and blurry. Complete removal of the artifacts would be nice for sure, but getting rid of 80% of them is also good enough really.
I’m going to cure the artifact menace by leveling them out, but that will require that I recover parts of the page from overleveling by using curves to change up the tones or by reducing the leveling strength on the overleveled parts of the page. Using the mask layer of the leveling adjustment layer, I brush mid-gray or dark gray (using the web colors only palette) to remove some but not all of the adjustment. The areas I targeted were the girl’s neckband in p1 and dark areas of her dress in p2 where the black leveling ended up darkening areas so much that details were lost in shadow. Other parts that need help are the very light tones that are present in p4 where the light white leveling still makes it very hard to see. However, I don’t reduce leveling on those areas because there’s a bunch of artifacts scattered throughout and I don’t want to bring those back. Instead, I whip out the burn tool (set to shadows to try to avoid bringing out the artifacts) to darken the patterns a bit to make them more visible.
Update:
I wrote the cleaning section in its entirety before being told of this handy anti-artifact method, so I'll add it here and leave the rest of the sections mostly untouched. It is, after all, how I did it for way too long until now...
For those who don't know the program name Waifu2x, it's an AI-trained program that runs on the nvidia GPU cuda cores to perform upscaling and/or denoising without drastic (but still present) losses of quality like older algorithms.
The details are all originally from this Google Doc, but I'll rephrase it here with my own cut down examples for direct comparison's sake since I will be using the images part of the guide .zip package.
I'm using one of the many versions of it from https://github.com/lltcggie/waifu2x-caffe/releases but there are other flavors out there for sure.
The first step is to run the raw files through Waifu2x at level 3 denoise, but don't overwrite the original files. The only other option I changed was to set the output to .png instead of .jpg since we're trying to get rid of artifacts with this method, not maintain them.
Next is to go back to photoshop and open the original, non-waifu'd file and paste in the waifu'd version as a new layer. I still perform the safety step of duplicating the background layer and hiding it as a force of habit.
Next step is to duplicate the Waifu layer twice. I'd rename the waifu layer to something more memorable instead of the default "Layer 1", rename the top waifu duplicate as something regarding the white levels, and rename the bottom waifu duplicate as something regarding the black levels. Hide the original waifu layer.
Proceed to perform a levels adjustment directly onto the whites layer first (shortcut CTRL + L) instead of creating an adjustment layer because we want this process to be intentionally destructive (that's why we have a few copies of it going). You'll be moving the midtones slider (middle gray slider) all the way to the right to make the image dark as hell and blow out the jpg artifacts into groups of grayish blocks. Now that you can visually see the jpg artifacts, you will lower the blacks level one by one until those blocks are gone in a satisfying manner. The same adage for standard leveling applies where you shouldn't overdo this; the change shouldn't typically be any higher than 15 (depends on the state of the raw, however).
For page 019 I dropped the whites level on the "whites" layer from 255 to 247. You're turning the grayish jpg artifacts to pure white so there is some leeway with going overboard (just be sure to check after everything's done that it doesn't rip away details). After setting the whites level, change the blending option (drop-down box in the layers window of all the various things like "Normal", "Dissolve", "Darken", "Multiply", etc.) to "Lighten" and then hide the layer. Be sure to check that it actually did what it was supposed to before hiding it.
I rose the blacks level from 0 to 10. It's a harder call here whether to raise or lower it more because the loss of detail was more recognizable here. I focused on the dress and whether the pure black conversion was going to overtake the dress details. Keep in mind that you will still perform a standard leveling adjustment layer even after this since this is just to help remove the jpg artifacts not to level the page. Set the blending option for this layer to "Darken".
The blending options, as described by Adobe themselves:
Lighten - Looks at the color information in each channel and selects the base or blend color, whichever is lighter, as the result color. Pixels darker than the blend color are replaced, and pixels lighter than the blend color do not change.
Darken - Looks at the color information in each channel and selects the base or blend color, whichever is darker, as the result color. Pixels lighter than the blend color are replaced, and pixels darker than the blend color do not change.
The majority of the color information on the two layers were shifted with the sliding of the midtones scale to the extreme ends. The lightened layer had a lot of the colors shifted super dark, but then the white level was shifted back a bit which caused the majority of the gray clouds that were the artifacts to become pure white while most of the rest of the image was untouched. Thus the now more lighter areas, set to lighten blending, take priority. The same is true of the darkened layer but in the opposite direction. A neat solution I wouldn't have imagined due to my lack of knowledge on proper blending option usage.
As the doc states, you'll want to keep the lighten layer above the darken layer. With this, there is a drastic reduction of the jpg artifacts without a large shift in quality for the rest of the image and without a ton of extra work that needs to be done by hand like I had written out before. The final image results for the pages with strong artifacting use this additional method of cleaning.
Now for redrawing p1 and p2 I’ll do the same deal as in the last panel on page 15 where I typeset first before filling in the blank spots. The patterns aren’t too tough, just irregular looking, but we can always save ourselves some work by doing this. However, keep in mind that if you ever need to move or adjust the typesetting for any reason (like after a QC pass) you end up doing more work trying to re-adjust everything so it’s not like there’s no downsides to this.
I did nothing too special for p1 and p2 redraws. Just repeated and varied sampling to create an approximation and some dodging/burning to adjust. The font I used for the sfx here was Twelve Ton Fishsticks. For p2 I chose to do the text covering on the bottom-right area. There is the option of doing it on the top-left area, but it’s easier to clone that zone and also the kicking action is originating from the bottom-right so it makes more sense to place it there. There was an instance where I copied a block of the background, flipped it horizontally, and then free transformed it to match a bit of the missing pattern for the bottom-right redraw due to lack of good cloneable areas.
I chose the font PP Handwriting for the zwoosh sound effect, Never Say Never for the tap bubble, ComixHeavy for the flap bubble, and Written in the Stars for p3b1small. Wild Words was used for the dialogue bubbles and CCZoinks using only caps for the emphasized text. With leading set to auto I find that I don’t like the amount of extra space the text gets between lines for Wild Words, so I changed the leading to font size + 2 (+4 for CCZoinks).
There’s nothing special about the typesetting for the rest of the page except for the heart in p1b1. There’s no standardized way for placing the heart so I’m just going to do it the way I normally do it by having it in-line and you can see if you like it aesthetically. I went with the period after the heart because I think it looks better than the heart after the period. Once the heart was in the position I liked, I linked the text and the heart layer together to make movement easier for when I was recentering the text in the bubble.
For p3b3 I centered the text as if the rest of the bubble was not cut off by the panel. This is not a hard rule of mine as there have been a good number of occasions that the placement for the text in that situation would have led to it looking weird. P4b1 I’ve seen people leave it uncleaned (and I’ve done it myself sometimes), however I’m now of the opinion of cleaning it out and typesetting the same thing back in except horizontally because that’s how we’re reading everything else. Lastly for p4b2 I chose again to not hyphenate the word and with some size reduction I fit the sentence in without making it too small.
This page was a general-purpose page with a little of everything. What it really tested for was your eye for detail preservation.
.
.
.
This certainly isn’t one of the worst pages with gutter (you can even get a glimpse of a harder one on the next page). The gutter refers to the dark and warped gap down the side of a raw, or down the center if the scan is double-paged, caused by the page(s) still being bound to the book. All of the other pages in this test .zip were from debinded books. Besides the handling of minor gutter, this page is mainly testing your ability to redraw hair because all mangaka artists LOVE covering their characters’s head and hair with text. You tend to learn some drawing patterns for hair after doing enough of them even without using a tablet. If the series you’re working with has way too many of these where the time consumed is astronomical, you can potentially do a half-step between white-boxing and full redraw by leaving the area within the text being whited out and then redrawing the surroundings.
Things to watch out for as you work on this page especially because it’s a magazine scan:
- There is no perfect place to choose to use as a straight line for straightening because where one location is straight, the others will not be due to the page being curved from the scan.
- Shading details will be weakened from white leveling away the magazine grayness. If you were to underlevel to preserve them you introduce a bunch of dust so I’d say either use curve adjustment instead and fiddle with the anchor points, mess with the mask layer on the leveling layer to bring back the detail, or work on your original copy layer with the burn tool to strengthen the details for post-leveling. Due to the strength of the leveling needed I’m not sure there’s much space to mess with the curve afterwards so I’m going to work on bringing back the details.
- There will be some leftover dust to clean after leveling (or ignore if you don’t care). If you used leveling to get rid of every single speck then you’ve overleveled.
- There’s minor bleeding in panel 1 and moderate bleeding in panel 2 to deal with.
- The hair redraw in panel 2 may look intimidating, but the pen tool (or standard brush strokes at low softness if you’re confident in your mouse control) is your absolute best friend here and there is at least an good amount of cloning area to work with. Also whiting out the text will help with the actual redrawing process and mentally perceiving what’s needed to fill the gaps. The cloning of the tones isn’t actually as bad as the magazine scan makes it seem at first. This is mainly a time sink so most of the difficulty is just cloning the tones into the missing areas.
- Due to the strength of the leveling layer you may want to redraw on a layer above the adjustment layer or change the brush color from pure black to a mid-level gray.
Fonts used: Set Fire to the Rain (p1b1) and the rest used Komika Text.
I debated whether to have the text in p6 in the open space on the boobage or over the redrawn areas; I settled on over the hair to avoid covering up that area despite there technically being nothing.
At this point after you complete this page, the other two pages don’t serve any more teaching opportunities. I did nothing different from those pages that I didn’t do from the first four. I’ll subject myself to the torture and complete them again anyway to provide a comparison for everyone to look at.
.
.
.
Welcome to redraw hell. While the redraws here aren’t exactly super tough, it’s just numerous to the point where you might end up spending enough time on this one page to hate it and curse out the artist for their anti-scanlator DRM. If you’re not going to redraw/translate sfx then the work is greatly lessened.
Panel 3 is by far the worst one and will require the most time and patience to draw art that isn’t yours. Because the colors are inverted you can’t completely erase the sfx before beginning the redraw. You may also notice that hands are a bit funky in their proportions. I don’t fix mistakes or weirdness made by the artist I just do the best that I can at replicating the intended art unless it makes it look too weird to after the redraw to leave it as is.
For the last panel it may be easier to repetitively clone the pattern until all of the space is filled in properly instead of placing the typesetting in first., that way the pattern can be more readily followed for a more consistent clone. Again, it’s relatively easier to see an improperly pattern than it is seeing minor mistakes or repetitions in the pattern when it’s surrounded by a bunch of other stuff that seems correct.
Panel 5 I took the liberty of widening the bubble which is too thin for horizontal placement and I’m not a huge fan of making the text vertical unless it’s for stylistic reasons like with the yelling in panel 1. Nothing from the art is being lost and the text isn’t overlapping the bubble border which I find rarely looks pleasant. I also took the liberty of filling in the white space underneath satisfied and then drawing my own blob of white to “fit” the text better despite the original only semi-conforming.
This page took a silly amount of time to satisfy my own personal requirements, but depending on the series it won’t be a common occurrence.
Fonts used: Mighty Zeo (dialogue), Minneapolis (shriek/yell/exclamation), Another (emphasis), Porky’s (squishing), Waiting for the Sunrise (p5b1), and Yellowjacket (“Satisfied”)
It was at this point (or a little bit earlier) that I knew that I either needed to go find more fonts or end my decision to use a unique font for every page. With only one page left, I decided to roll with what I had left. These are partially font demonstrations and not necessarily exactly how I would use them!
.
.
.
It’s clear from the outset that there’s only one thing to really look out for on this relatively easy last page, and that’s panel 4’s redraw. There are some of the same things to watch out for as from previous pages with regards to stuff like the jpg artifacts and loss of detail from leveling.
The point of this page wasn’t only to check the redraw capability, but to see how people treat pages that require less time from them. Like the first page from the tutorial, the simplicity of the page leads to minor mistakes sticking out more easily than a busy page like the Otoboku one. Unlike the first page that had noticeable bleeding if left untreated, this page was one of the worst offenders for poor efforts simply because it’s mostly a basic page. While last numerically, it was actually one of the pages I would check first (alongside 03_177) to see how much time I should spend checking the other pages to determine if the applicant passed a threshold of editing competency. Not everyone will work on pages in order from alphanumeric order like we’ve done here, but I feel it’s a nice way to bookend the editing guide section.
Here’s the state I left the redraw at before deciding to put text over it. It’s certainly not perfect, but it’s close enough once you slap text over it. I could have possibly saved a chunk of time had I placed the text first.
Even if you’re not confident and want to slap text over it and fill in the gaps from there, I believe it’s a good skill to have to perform these redraws of character models if you plan on scanlating for any length of time and are not making speed scanlations where quality goes out the window.
The font used was KG Behind These Hazel Eyes.
.
.
.
MISC:
Curves vs Leveling
I mentioned way back that you could optionally use curves to do the same thing as leveling plus a little more. The reason I didn’t advocate for it was because it introduces a perhaps unnecessary layer of complexity that and because it’s not always better.
Most of the settings are similar to the leveling adjustment settings with the exception of the ability to add anchor points to adjust the line into a curve. With proper handling this can potentially alleviate some of the effects of overleveling without going into the mask layer to bring back lost detail. It’s also very easy to mishandle the anchoring and just end up wasting more time, which I didn’t want to add on top of all the other things I wanted to write while you’re in the midst of things. There is also one key difference in the settings between curves and leveling and that is leveling goes on a scale from 0-255 while curves work from a scale of 0-100. There is a finer level of control granted to leveling that I did not acknowledge in my last guide from a decade ago (for anyone who’s bothered to read it at this point with no images to work from). If you use curves for even simple leveling you may end up slightly overleveling purely from the fact that every point up or down on a curve scale is equal to 2.5 points on the leveling scale. I encourage everyone to at least try it once or twice to see how it works since it could save some brushing action on the mask layer if used effectively.
To be included at a later date: Raster vs Vector manipulation (introduction to Illustrator or other vector-based editors)
last edited at Dec 13, 2021 3:48AM