Most logos are designed for a surface that holds still. A sign, a business card, the corner of a website, a product label — each one gives the mark all the time in the world. Video does the opposite. It gives the mark a fraction of a second, in a frame that is often a few centimetres wide and frequently moving, and then it moves on. The logos that survive that transition are rarely the ones that look best on a billboard. They are the ones that were drawn with a small frame and a brief glance in mind.
That is the whole brief behind a logo for video. It is not a smaller version of the main mark, and it is not the main mark with an animation bolted on afterward. It is a sibling drawing built for the conditions video imposes, and getting it right takes about an hour if you know which decisions matter. Here is what those decisions are.
Why the Main Logo Usually Fails on Video
Start with the arithmetic, because it explains almost every failure. A mark that works on a 300-pixel website header is being asked to hold up inside a frame where the whole width might be 1080 pixels, but where the mark itself occupies perhaps a fifth of that. Then it gets compressed, viewed on a phone in daylight, and scrolled past. Every element that depended on fine detail is now gone.
Three specific things break, in this order.
Thin strokes disappear first. A hairline rule or a delicate serif that reads as elegant at large sizes turns into a faint grey smudge at video scale, and compression softens it further. The mark keeps its general shape and loses its character, which is arguably worse than losing it entirely.
Detailed sub-elements go next. A crest with a dozen internal lines, a wordmark with a finely drawn ampersand, an icon with internal texture — all of these survive as vague noise. Video encoders are built to throw away exactly the kind of information those details are made of.
The wordmark stops being readable last, and this is the one people fight. Below a certain size, letters stop being letters. They become rhythm and texture, which means a viewer can no longer read your name off the frame — only recognise that there was some text there. If recognition is the job, the wordmark is doing less work than it costs in space.
There is a mechanical reason compression hits these details hardest. Video codecs allocate bits to what changes between frames, and fine static detail is expensive to preserve across a whole second of footage. The practical consequences are documented in MDN's guide to video codecs, which is worth skimming before deciding how much detail a small mark can carry.
The Rule That Saves the Most Rework: Build the Small Frame First
The instinct is to design the biggest version and scale down, because a big canvas shows more detail and feels like more room to work. It is the wrong direction for video, and it costs an afternoon every time.
Design the smallest placement you will ever ship, then scale up. That surface is usually the avatar — a circular crop, displayed at a fraction of a screen width, sitting beside a handle in a comment thread. If the mark holds there, it holds at every larger size, because every other placement is the same drawing with more pixels. If it fails there, you have learned that in an hour instead of after building and exporting a full animation that turns out to be illegible where it matters most.
Practical translation: put the mark inside the inscribed circle of a square, keep it in roughly the middle seventy percent, and check it at actual display size on an actual phone. Not in a design tool at 400 percent, and not on a desktop monitor. Those two views are the reason good marks ship broken.
What Actually Changes Between the Static and Video Versions
Once you accept the video version is a sibling rather than a copy, the decisions get concrete. These are the ones that pay.
Drop the strapline. Anything that reads after the name — a tagline, a founding year, a city — is gone in the video version. It cannot be read in the time available, and it takes space the mark needs.
Increase stroke weight rather than decreasing it. Video softens edges through scaling and compression, so a line that looks correct on a static file at the same size will look thinner on screen. Compensate in the source, not in the export.
Simplify the symbol to one idea. If the main mark carries a symbol and a wordmark and a strapline, the video version keeps the symbol. If the symbol itself has a primary shape and a secondary shape, the video version keeps the primary.
Choose a shape that survives a circular crop. Wide horizontal lockups lose their ends. Marks with elements pinned to corners lose those elements. A compact, tall or diagonal composition inside a circle loses nothing.
Lose the thin outline. Fine strokes are the first thing compression removes, and an outline that vanishes leaves the mark looking unfinished rather than minimal.
Enough About Sizes: The Logo for Video Is Usually Seen Twice
There is a surface where this whole problem is already solved, and it is worth studying rather than inventing from scratch. Broadcaster branding practice has spent decades working out what happens when a logo has to survive being shrunk, cropped, replicated across dozens of contexts, and shown against arbitrary backgrounds. The conventions that came out of it are explicit, and one of them is directly relevant: a mark should arrive cleanly rather than decorated.
YouTube publishes its own channel art and avatar guidance for much the same reason, and it is unusually specific about sizes and safe areas. The short version from YouTube's channel branding documentation is that an avatar is displayed at 98 by 98 pixels and that additional decoration — borders, shadows, frames — is not recommended. A mark that needs a border to work on video does not work on video.
For frame proportions, the same source gives the working numbers: YouTube's aspect ratio and resolution guidance establishes 9:16 vertical, 1:1 square and 16:9 widescreen as the shapes that matter, and recommends 1080 by 1920 for vertical. That is the export set a video logo actually needs: three renditions of one drawing, plus a still for avatars and thumbnails.
Notice that none of these sources describe a single master file. Every platform expects a set, sized differently for each placement, and treats anything else as a compromise. Building that set deliberately is the difference between a mark that looks right everywhere and a mark that looks approximately right everywhere.
Motion Changes What the Drawing Has to Do
A static mark has one job: be recognised. A video mark has two — be recognised, and complete an action inside a time budget. That second job is what makes a logo for video a different design problem rather than a smaller one.
The action is short. An entrance, a hold, an exit. Under a second for the arrival is the working target, because a viewer scrolling a feed is already deciding whether to keep watching before the mark has finished appearing.
Three design consequences follow.
The drawing has to read as finished at the moment it arrives, not partway through. If the mark is built from pieces that assemble — letters flying in, a symbol rotating into place — the viewer only sees the complete brand for the final fraction of the sequence. The mid-states have to be recognisable, or the animation is spending the recognition budget on the process instead of the result.
Elements that move need enough separation to move into. A tightly packed lockup gives an animation nowhere to travel, so an entrance has to be a fade or a scale from nothing — both of which read as generic. Loosening the composition slightly in the video version creates room for a deliberate entrance.
The end state has to be a clean hold. If the sequence finishes on a slightly rotated or partially transparent frame, that becomes the image a viewer remembers. The last frame should be the mark, presented properly, held long enough to register.
There is a well-documented argument for keeping all of this restrained. Nielsen Norman Group's research on animation in interfaces draws a clear line between motion that communicates a state change and motion that exists for its own sake, and finds that the gratuitous kind costs more attention than it returns. A logo entrance is a state change — content is beginning — so it qualifies. A logo entrance with a spin, a bounce and a particle burst is decoration, and it spends the exact resource the opener was supposed to buy.
A Working Sequence, Start to Finish
Here is the order that avoids rework, based on the sequence that causes the fewest problems.
One. Export the mark as an SVG, or as a transparent PNG at the largest resolution available. Confirm the transparency by viewing the file against a dark background rather than trusting the export dialog. This step derails more projects than any other, because a JPEG lifted from a slide deck carries an invisible white rectangle that becomes glaringly obvious once the background is not white.
Two. Build the circular avatar crop first, and check it on a phone. This is the constraint the rest of the set inherits.
Three. If the wordmark has stopped being legible in that crop, make the decision now: drop it from the video version and keep the symbol only. Deciding this after an animation exists means rebuilding the animation.
Four. Design the entrance as a one-second idea, then export the three ratios from one source so timing does not drift between them.
Five. Export a still frame from the clean hold at the end of the sequence, and keep it as the avatar and thumbnail candidate.
Six. Watch the result muted, at feed size, on a phone. If the entrance still reads as deliberate with no sound, the timing is doing its job. You can see this sequence in practice on a <a href="/logo-animation-templates/logo-video-maker">logo video maker</a> page, or compare what different generated approaches produce on the same file in our walkthrough of <a href="/blog/new-video-logo">building a new video logo from scratch</a>.
Mistakes That Cost the Most Time
Four of these account for most of the rework people report.
Starting from the largest version and scaling down. It feels efficient and it hides the failures until the worst possible moment.
Reusing the web export. A logo exported for a site header is often flattened, sometimes on a white background, and rarely at a resolution that survives a vertical canvas. Re-export from the source file.
Treating the animation as the design. Motion cannot rescue a composition that does not read in a small frame — it can only draw attention to the problem for slightly longer.
The opposite error also exists: spending weeks on a version of the mark so minimal it stops being distinctive. A mark optimised entirely for a 40-pixel circle can lose the one shape that made it recognisable.
FAQ
Can I use my existing logo on video?
Usually, but not the existing file. Take the source artwork and export a version for video specifically: heavier strokes, no strapline, the symbol prioritised, and confirmed transparency. The drawing can be the same one; the export has to be different.
What size should a logo be in a video?
There is no single number, because it depends on the placement. The workable approach is to design against the smallest surface — the circular avatar — and scale up from there, then export 9:16, 1:1 and 16:9 renditions from one source so all three agree.
Should the logo appear at the start or the end of a video?
Start if the mark is the reason the viewer is watching, which is the case for branded content and openers. End if the viewer came for the content and the mark is a signature. Putting it in both places is common and usually a mistake unless the piece is under ten seconds and loops.
Do I need the wordmark in the video version?
Check it at display size before deciding. If the letters have stopped being readable, the wordmark is occupying space without communicating a name — and on most platforms the account name is printed as text immediately beside the mark anyway, which makes it redundant as well as unreadable.
How many versions of a logo does a video channel need?
Four working files: a circular still for avatars, a square still for thumbnails and feed placements, a vertical animation, and a widescreen animation. All four come from one drawing, so the cost is in the export pass rather than in the design.
The Bottom Line
Take your logo file and open it at the size it will actually be seen — a small circle on a phone, in daylight, in a feed that is already moving. If you cannot tell what it is, the video version does not exist yet, no matter how good the main mark looks. Fix the drawing before you touch the animation, because motion multiplies whatever the composition already does: it makes a clear mark clearer and a crowded one more confusing. The honest test takes two minutes, and it is the only one that predicts how the logo will perform where it will actually live.