Homemade Hanpen as Fish Clouds
Updated: Oct 2
Mass-produced hanpen is built to be forgettable — bland on purpose, a texture-forward filler for stews and soups that's sold pre-cooked and made only to be reheated. Homemade is a different food entirely. It tastes like the fish you used, it's cloud-light in a way that can't survive a second cooking, and it's three ingredients you can have in a food processor in minutes. I got into this because it's cheap, fast, and fun — and it turned out to be a good way to add variety to a lean-protein rotation, with more room to play with fish, seafood, and presentation than I expected from something this simple.
People ask why, and more than once — the last hanpen video drew a string of "why bother?" comments. It's a question I ask myself too, every time I'm turning a recipe over in my head. Which moves with an ingredient actually make sense, and which don't.
But I look at food through curiosity and through what it does for the person eating it, not through efficiency. What happens if I do this to these ingredients? How do they behave together? Which one is pulling which string in the final result, and how does it actually feel in the mouth? Food isn't fuel shoveled into a furnace for me, and it isn't a chore to get through as fast as possible. Hanpen is an extra texture, a way to turn inexpensive, not especially interesting fish into something worth remembering — and, in a professional kitchen, a legitimate way to turn fish trim into something plate-worthy. That last part only holds if you actually have trim lying around, though, which mostly means working cooks and the more serious home cooks who break down whole fish themselves.
What is hanpen
Hanpen is a steamed Japanese fish cake from the Edo period, likely named for a Suruga chef called Hanpei — or, arguably, for its half-moon shape. It's white fish, egg white, grated nagaimo (mountain yam), and kombu dashi, and it's designed to taste like almost nothing — its job is to soak up other flavors. Today it's mass-produced from frozen surimi on industrial mixers that whip air into the paste under refrigeration, then steamed at 80–90°C and sold pre-cooked, ready to reheat (more on production).
The food science — and what actually matters
Two things are happening simultaneously and homemade success depends on both.
Protein gelation — salt solubilizes the myofibril proteins (actomyosin) in the fish, and heat sets them into a cohesive gel network, which is what holds the shape once steamed;
Air incorporation — whipped egg white plus the mucilage from grated nagaimo trap and stabilize air bubbles in the paste, giving hanpen its signature spongy, almost foam-like crumb.
At home, a food processor stands in for the industrial mixer, but you lose the precise temperature control.
That first step — fish and salt, alone, before anything else — is the one every surimi product actually depends on. Commercial kamaboko (type of surimi) production grinds minced fish with 2–3% salt for 30–40 minutes at near 0–10°C before starch or seasoning goes in, because the salt needs direct, undiluted contact with the muscle tissue: the ions disrupt the protein's internal ionic bonds and pull myosin into solution, which is what later sets into the gel network that gives the cake its structure (salt-soluble protein extraction). Add starch or egg white too early, industrially, and that contact gets diluted and the gel weakens.
I tested this at home two ways — fish and salt processed first, then yam, then egg white; versus everything in the jar at once — and couldn't tell the difference. I'm leaving that as an open question. If you try this at home, test both ways and see if you can tell them apart. I'd genuinely like to know what you find.
Fish for homemade hanpen
I have tried lean white fish (sole, halibut, lingcod) alongside fattier, pigmented fish (steelhead, salmon). The differences are real but subtle. Fattier fish shifts color (pink-tinted) and brings its own flavor forward, but fat content doesn't meaningfully change the texture or how well a piped shape holds — the gel forms and sets about the same either way. It was interesting to combine both types of fish in one bite.
Finding the sweet spot for nagaimo by weight
I expected yam vs. potato to be the interesting variable — it isn't. Swapped potato (Yukon Gold variety) in for nagaimo and the textural difference was minor. What matters is how much starchy binder goes in.
Recipe:
300 g fish fillet
½ tsp salt
¼ tsp sugar (optional)
75 g nagaimo (Japanese mountain yam) or Yukon Gold potato, grated
30 g egg white
Nagaimo / Potato | Result |
65g | Firm, bouncy |
75g | Balanced per my taste |
85g | Fluffy, spongy |
95-100g | Even lighter, but won't hold a piped shape through steaming |
A 20g swing across 300g of fish is the difference between a cake that pipes clean and one that slumps in the steamer — worth calling out as the single most important number in the whole process.
Piping and extrusion
That was my favorite part! I tried different types and sizes of piping tips, looking for the best shape — one that would make sense from a few different angles at once: size, simplicity, aesthetics, and logistics. Really large piping tips won on all of them. They shape a reasonably sized fish cake in just one or two squeezes. My favorite round tip is Sultan. The shape suggests to use the cavity in the center for toppings. Another shape that I liked and repeated a few times was the one mad with extra large Basketweave piping tip. It produces a long wavy strip that can nicely hold a sauce and be decorated with assorted toppings.
Thickness of the shape you choose is the variable worth flagging, because it drives steaming time. Heat has to travel from the outside of the cake to the center before the gel fully sets. A 1/3" thick wave shape made with Basketweave is ready in about 5 minutes, while the larger round shapes made with Sultan need 8–10.
Sauces and toppings
My first steamed fish cakes were served with just melted butter — and that was enough to enjoy them! When I made the "Sultans," I added a bit of salmon roe in the center and dusted the cakes with aonori (dry seaweed with an aroma similar to fresh cucumber). Furikake works very well as a topping here too — convenient, logical for seafood, and it adds color and texture.
Any French butter- or cream-based sauce elevates this dish and moves it away from Asian cuisine — beurre blanc, hollandaise, béarnaise, béchamel, suprême. Toppings like capers, caviar, microgreens, and edible flowers finish the plate. And it's not a humble hanpen anymore.
Where I stop, for now
I had an idea to pipe the paste straight onto the serving plate, arrange raw steelhead and shrimp on top, steam the whole plate, serve it as-is. I didn't like the result. The shrimp curled and deformed in a way I didn't want, and the technique just didn't deliver the clean, composed look I was picturing. So that specific idea — cook and serve on the same plate — doesn't make the cut. I'm noting it here instead of quietly dropping it, because a blog about experiments should own the ones that fail.
That would have been a tidy place to end experimenting with homemade hanpen, except for one problem. I had video, but not a single usable photo for the blog. So I went back into the kitchen to make hanpen again, purely to get stills for this post. What followed was three failures in a row, and they turned out to be the most useful part of the whole project.
First attempt. I grabbed whatever potatoes were on hand — small, young ones, some pink-skinned, some Yukon Gold — and got a watery mess instead of a paste. Young, waxy potatoes carry more water and less starch than a mature baking potato, and that difference is enough to sink the recipe. Section 4's finding — that potato works about as well as nagaimo — only holds for a starchy, mature potato. Swap in anything waxier and you're not dealing with a minor variable anymore.
Second attempt. I bought frozen cod, thawed it, and had a bad feeling before I'd even started — the fillet was shedding water constantly. It was right to worry. The result was another watery mass, this time with visible flakes of fish protein instead of a smooth paste. Freezing and thawing quality turned out to matter as much as the fish species: damaged, waterlogged flesh doesn't have the structural integrity to gel properly, no matter how good the ratios are.
Third attempt, I genuinely don't have an explanation for. The mixture came out looking like meringue, or shaving foam — airy in a way that had nothing to do with hanpen's usual spongy bite.
After three failures back to back, I stopped for a while — about six weeks, as it turned out. I'd gotten complacent. Early on, nothing I tried seemed to break the recipe, so I assumed there was a wide margin before I'd hit an actual wall. There wasn't. The food science I'd written about in section 2 wasn't decoration — it was the thing I should have been paying closer attention to the whole time.
Yesterday I bought a flounder fillet that looked right, and today, with nagaimo instead of potato (no more guessing), it worked. That's where these photos come from, and that's where I'm stopping — not with the elaborate combination plate I'd planned, but with carefully composed dish that reminded me why the fundamentals matter in the first place.
Comments